ID	Accession_number	Authors	Gene	Gene_ID	Antisense_21mer	Sense_19mer	5.end	3' end	Sec.dG	Hsieh	Amarzguioui	GC stretch	Whole ∆G	Katoh	Takasaki	Reynolds	Ui-Tei	%GC	i-score	%Inhibition	Technology	Cell	Concentration	Hours	Title	PMID	Year	Abstract	Journal	Pubdate
Si0001	NM_012249	Huesken	TC10	23433	CUAAUAUGUUAAUUGAUUUau	AAAUCAAUUAACAUAUUAG	-0.9	-2.1	2.1	0	-1	1	-24.6	95.4	-1.3	6	II	15.8	57	46.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0002	NM_012249	Huesken	TC10	23433	AAUAUGUUAAUUGAUUUAUac	AUAAAUCAAUUAACAUAUU	-1.1	-0.9	2.5	2	1	1	-23.6	104.2	-1	9	II	10.5	65.3	38.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0003	NM_012249	Huesken	TC10	23433	GAUUUAUACAAUUCCUUUCaa	GAAAGGAAUUGUAUAAAUC	-2.4	-2.4	2.3	2	1	2	-28.2	102.6	12.7	7	II	26.3	61.5	51.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0004	NM_012249	Huesken	TC10	23433	CAAUUCCUUUCAAUUUUAUcu	AUAAAAUUGAAAGGAAUUG	-1.1	-2.1	1.3	-1	-1	2	-26.2	87.4	-8.6	5	II	21.1	44.5	36.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0005	NM_012249	Huesken	TC10	23433	CAGACCAAAAUUAAAUAAGaa	CUUAUUUAAUUUUGGUCUG	-2.1	-2.1	2.8	-1	-1	2	-27.4	79.6	-1.6	4	II	26.3	56.4	52.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0006	NM_012249	Huesken	TC10	23433	AGACCAAAAUUAAAUAAGAaa	UCUUAUUUAAUUUUGGUCU	-2.4	-2.1	2.8	2	0	2	-27.7	65.6	-6.3	4	II	21.1	51.3	44.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0007	NM_012249	Huesken	TC10	23433	ACCAAAAUUAAAUAAGAAAgu	UUUCUUAUUUAAUUUUGGU	-0.9	-2.2	2.8	1	-2	2	-25	79.6	1.3	6	II	15.8	48.6	44.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0008	NM_012249	Huesken	TC10	23433	CAAAAUUAAAUAAGAAAGUua	ACUUUCUUAUUUAAUUUUG	-2.2	-2.1	3.7	-1	-1	1	-23.8	77.8	-5.9	6	II	15.8	59.3	43.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0009	NM_012249	Huesken	TC10	23433	UAAGAAAGUUACAUAAGAUuc	AUCUUAUGUAACUUUCUUA	-1.1	-1.3	3.5	0	3	1	-28.2	78.1	-1.6	7	II	21.1	64.2	59.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0010	NM_012249	Huesken	TC10	23433	AAGUUACAUAAGAUUCCAUuu	AUGGAAUCUUAUGUAACUU	-1.1	-0.9	2.3	1	2	2	-30.4	81.9	-1.9	6	II	26.3	54.2	51.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0011	NM_012249	Huesken	TC10	23433	ACAUAAGAUUCCAUUUGAGca	CUCAAAUGGAAUCUUAUGU	-2.1	-2.2	1.4	1	4	2	-31.4	68.1	-2.3	6	Ia	31.6	52	55.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0012	NM_012249	Huesken	TC10	23433	AAGAUUCCAUUUGAGCAUAca	UAUGCUCAAAUGGAAUCUU	-1.3	-0.9	1.4	1	-1	2	-32.6	71.6	-0.7	5	II	31.6	50.8	54.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0013	NM_012249	Huesken	TC10	23433	CCAUUUGAGCAUACAUAAGgc	CUUAUGUAUGCUCAAAUGG	-2.1	-3.3	1.2	0	-1	2	-32.5	69	0.7	4	II	36.8	46	44	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0014	NM_012249	Huesken	TC10	23433	CAUUUGAGCAUACAUAAGGcc	CCUUAUGUAUGCUCAAAUG	-3.3	-2.1	1.2	0	1	2	-32.5	74.1	0.4	6	II	36.8	63.2	65.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0015	NM_012249	Huesken	TC10	23433	AUAAGGCCAUGAUACUUUAau	UAAAGUAUCAUGGCCUUAU	-1.3	-1.1	1.3	1	0	4	-32.9	70.1	-3.8	6	II	31.6	51.7	75.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0016	NM_012249	Huesken	TC10	23433	GCCAUGAUACUUUAAUGUGaa	CACAUUAAAGUAUCAUGGC	-2.1	-3.4	1.2	0	0	3	-32.6	69	3	3	II	36.8	45.2	62.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0017	NM_012249	Huesken	TC10	23433	UUAAUGUGAACCACCAUUUcu	AAAUGGUGGUUCACAUUAA	-0.9	-0.9	-2.4	1	1	2	-32	85.4	-11.3	9	II	31.6	76.4	85.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0018	NM_012249	Huesken	TC10	23433	UGUGAACCACCAUUUCUUGga	CAAGAAAUGGUGGUUCACA	-2.1	-2.1	-0.3	0	3	2	-35.3	70.6	2.7	6	Ib	42.1	58.9	84.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0019	NM_012249	Huesken	TC10	23433	GAACCACCAUUUCUUGGAAga	UUCCAAGAAAUGGUGGUUC	-0.9	-2.4	-0.8	1	0	2	-35.5	58.4	4	2	III	42.1	27	38.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0020	NM_012249	Huesken	TC10	23433	AUUUCUUGGAAGAAAGAAGac	CUUCUUUCUUCCAAGAAAU	-2.1	-1.1	-1.4	3	3	2	-30.8	85.3	-0.3	6	Ia	31.6	66.7	53.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0021	NM_012249	Huesken	TC10	23433	UGGAAGAAAGAAGACAUCCaa	GGAUGUCUUCUUUCUUCCA	-3.3	-2.1	-1.8	0	2	2	-36	64.9	12.4	6	Ib	42.1	74.2	83.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0022	NM_012249	Huesken	TC10	23433	GGAAGAAAGAAGACAUCCAaa	UGGAUGUCUUCUUUCUUCC	-2.1	-3.3	-0.1	1	-1	2	-36	50.5	-4	4	II	42.1	40.7	68.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0023	NM_012249	Huesken	TC10	23433	GAAAGAAGACAUCCAAAUGuc	CAUUUGGAUGUCUUCUUUC	-2.1	-2.4	0.2	0	1	2	-32.3	66.7	7.7	4	II	36.8	50.6	73.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0024	NM_012249	Huesken	TC10	23433	AAGACAUCCAAAUGUCCGAuu	UCGGACAUUUGGAUGUCUU	-2.4	-0.9	-4.7	2	1	3	-36.3	68.9	-1.5	5	II	42.1	53.8	83	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0025	NM_012249	Huesken	TC10	23433	CAUCCAAAUGUCCGAUUCAga	UGAAUCGGACAUUUGGAUG	-2.1	-2.1	0.8	-1	-2	3	-35.2	54.7	-5.7	4	II	42.1	41.2	38.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0026	NM_012249	Huesken	TC10	23433	UUCCUGGCCAGUCAUCCAGua	CUGGAUGACUGGCCAGGAA	-2.1	-0.9	-4.5	2	3	4	-42.7	63.9	5.4	5	II	57.9	65.3	75.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0027	NM_012249	Huesken	TC10	23433	CCUGGCCAGUCAUCCAGUAga	UACUGGAUGACUGGCCAGG	-1.3	-3.3	-5	0	-3	4	-42.9	34.8	-8.1	-1	III	57.9	25.2	55.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0028	NM_012249	Huesken	TC10	23433	AGUCAUCCAGUAGACUCUCuc	GAGAGUCUACUGGAUGACU	-2.4	-2.1	-3	2	3	2	-39	52.8	10.1	3	Ib	47.4	55.6	77.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0029	NM_012249	Huesken	TC10	23433	AGACUCUCUCCACUCUUCAau	UGAAGAGUGGAGAGAGUCU	-2.1	-2.1	1.2	3	1	2	-39.8	66	-1.4	6	II	47.4	52.6	70.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0030	NM_012249	Huesken	TC10	23433	GGAAGGUGAUGCUUAUAUUuu	AAUAUAAGCAUCACCUUCC	-0.9	-3.3	1.9	1	-1	2	-34	66.1	-4	3	III	36.8	38	51.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0031	NM_012249	Huesken	TC10	23433	UCUACAAAGCUCACAUACAac	UGUAUGUGAGCUUUGUAGA	-2.1	-2.4	1.2	1	0	2	-35	65.7	-8.7	6	II	36.8	57.8	75.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0032	NM_012249	Huesken	TC10	23433	CAACAAGACAGUACAUCCUag	AGGAUGUACUGUCUUGUUG	-2.1	-2.1	-1.8	1	0	2	-35.8	58.6	-2.8	4	II	42.1	51.4	81.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0033	NM_012249	Huesken	TC10	23433	ACAGUACAUCCUAGAUUUGug	CAAAUCUAGGAUGUACUGU	-2.1	-2.2	-0.3	0	2	2	-33.9	58.5	-1.9	5	Ib	36.8	49.5	79.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0034	NM_012249	Huesken	TC10	23433	UCCUAGAUUUGUGACUGAUau	AUCAGUCACAAAUCUAGGA	-1.1	-2.4	1.5	0	0	2	-35.2	69.7	1.8	6	II	36.8	50.2	75.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0035	NM_012249	Huesken	TC10	23433	UUUGUGACUGAUAUGAGCAuu	UGCUCAUAUCAGUCACAAA	-2.1	-0.9	0.9	0	3	2	-35	81.4	1.3	7	II	36.8	71.6	83.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0036	NM_012249	Huesken	TC10	23433	UGUGACUGAUAUGAGCAUUua	AAUGCUCAUAUCAGUCACA	-0.9	-2.1	0.9	1	0	2	-35.2	71.3	4.1	6	II	36.8	56	76	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0037	NM_012249	Huesken	TC10	23433	GCAUUUAAGGCUGUCAUUUuc	AAAUGACAGCCUUAAAUGC	-0.9	-3.4	1.5	1	-1	3	-33.2	70.1	-1	5	II	36.8	44.3	42.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0038	NM_012249	Huesken	TC10	23433	UUUAAGGCUGUCAUUUUCAag	UGAAAAUGACAGCCUUAAA	-2.1	-0.9	2.4	0	2	3	-32	81.7	-3.7	9	II	31.6	71.2	78.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0039	NM_012249	Huesken	TC10	23433	UUAAGGCUGUCAUUUUCAAgu	UUGAAAAUGACAGCCUUAA	-0.9	-0.9	2.1	1	2	3	-32	76.1	-3.8	6	II	31.6	55.5	87.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0040	NM_012249	Huesken	TC10	23433	AAGUAUAAAAGUUUAGUGUuc	ACACUAAACUUUUAUACUU	-2.2	-0.9	2.9	1	2	1	-27.6	99.4	4.1	8	II	21.1	65.3	76.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0041	NM_012249	Huesken	TC10	23433	UAAAAGUUUAGUGUUCCAUua	AUGGAACACUAAACUUUUA	-1.1	-1.3	0.8	1	3	2	-29.8	92.4	4.1	9	II	26.3	70.5	82.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0042	NM_012249	Huesken	TC10	23433	GUGUUCCAUUACCCAAUCUgu	AGAUUGGGUAAUGGAACAC	-2.1	-2.2	1	0	-1	3	-35.9	69.2	-1.6	4	III	42.1	43.2	72	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0043	NM_012249	Huesken	TC10	23433	UGUUCCAUUACCCAAUCUGug	CAGAUUGGGUAAUGGAACA	-2.1	-2.1	1	0	3	3	-35.8	80.3	-2.3	7	Ib	42.1	67.6	86.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0044	NM_012249	Huesken	TC10	23433	GUUCCAUUACCCAAUCUGUgc	ACAGAUUGGGUAAUGGAAC	-2.2	-2.2	1.5	2	0	3	-35.9	65.3	-11.3	4	II	42.1	45.1	47.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0045	NM_012249	Huesken	TC10	23433	UGUGCAGUAGACAUAGCUAua	UAGCUAUGUCUACUGCACA	-1.3	-2.1	-0.8	0	1	2	-37.6	63.8	-4.1	5	II	42.1	52	86.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0046	NM_012249	Huesken	TC10	23433	UGCAGUAGACAUAGCUAUAag	UAUAGCUAUGUCUACUGCA	-1.3	-2.1	-0.8	-1	-2	2	-35.7	67.8	-1	5	II	36.8	50.5	77.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0047	NM_012249	Huesken	TC10	23433	UAGCUAUAAGCUCAAGUCAug	UGACUUGAGCUUAUAGCUA	-2.1	-1.3	-2.2	2	0	2	-35.3	73.6	-0.7	6	II	36.8	62.6	84.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0048	NM_012249	Huesken	TC10	23433	AUAAGCUCAAGUCAUGUGAau	UCACAUGACUUGAGCUUAU	-2.4	-1.1	0.8	2	1	2	-35	80.5	1.7	8	II	36.8	59.9	87	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0049	NM_012249	Huesken	TC10	23433	CUCAAGUCAUGUGAAUUAAca	UUAAUUCACAUGACUUGAG	-0.9	-2.1	0.7	0	-4	1	-31.3	69.2	0	5	II	31.6	37.5	54.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0050	NM_012249	Huesken	TC10	23433	UAUAAUUUGGGCAACAGCAag	UGCUGUUGCCCAAAUUAUA	-2.1	-1.3	-0.6	2	2	4	-34.5	79.6	-6	8	II	36.8	72.2	85.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0051	NM_012249	Huesken	TC10	23433	AUAAUUUGGGCAACAGCAAgu	UUGCUGUUGCCCAAAUUAU	-0.9	-1.1	-0.6	3	1	4	-34.1	65	-8.7	7	II	36.8	56	83.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0052	NM_012249	Huesken	TC10	23433	GCAACAGCAAGUUUAAAUGua	CAUUUAAACUUGCUGUUGC	-2.1	-3.4	0	0	1	2	-31.6	74.9	5.4	4	II	36.8	47.8	74.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0053	NM_012249	Huesken	TC10	23433	ACAGCAAGUUUAAAUGUAGga	CUACAUUUAAACUUGCUGU	-2.1	-2.2	1.5	1	2	2	-30.8	61.6	-2.6	5	Ib	31.6	55.4	82.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0054	NM_012249	Huesken	TC10	23433	AUUGUUUCUCUAAAAUGCAuu	UGCAUUUUAGAGAAACAAU	-2.1	-1.1	2.2	2	2	2	-29.8	77.4	-6.3	6	II	26.3	65.2	84.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0055	NM_012249	Huesken	TC10	23433	AUGCAUUAGGUUGUUCACAau	UGUGAACAACCUAAUGCAU	-2.1	-1.1	-0.2	4	1	2	-34.5	69.7	4.3	5	II	36.8	59.7	75.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0056	NM_012249	Huesken	TC10	23433	UGCAUUAGGUUGUUCACAAug	UUGUGAACAACCUAAUGCA	-0.9	-2.1	-0.8	1	0	2	-34.3	76.2	-1.4	6	II	36.8	54.5	83	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0057	NM_012249	Huesken	TC10	23433	UAACACCAUUAUCUGAUGCau	GCAUCAGAUAAUGGUGUUA	-3.4	-1.3	-4.7	1	5	2	-34.1	85.5	20.2	6	Ib	36.8	71.6	92.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0058	NM_012249	Huesken	TC10	23433	AACACCAUUAUCUGAUGCAuc	UGCAUCAGAUAAUGGUGUU	-2.1	-0.9	-4.9	0	1	2	-34.9	74	-6.1	5	II	36.8	55.2	61.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0059	NM_012249	Huesken	TC10	23433	CAUUAUCUGAUGCAUCCAUua	AUGGAUGCAUCAGAUAAUG	-1.1	-2.1	1.2	0	-1	2	-34.2	70.3	-0.9	5	II	36.8	44.9	82.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0060	NM_012249	Huesken	TC10	23433	UAUCUGAUGCAUCCAUUAUcu	AUAAUGGAUGCAUCAGAUA	-1.1	-1.3	1.1	2	1	2	-33.4	87	3.8	8	II	31.6	61.3	80.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0061	NM_012249	Huesken	TC10	23433	CUGAUGCAUCCAUUAUCUGau	CAGAUAAUGGAUGCAUCAG	-2.1	-2.1	0.7	-1	1	2	-35.2	65.1	-2	3	II	42.1	51.3	80.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0062	NM_012249	Huesken	TC10	23433	AUCCAUUAUCUGAUGCAUGgc	CAUGCAUCAGAUAAUGGAU	-2.1	-1.1	0.3	3	3	2	-34.2	73.3	2.4	5	Ia	36.8	66.5	72.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0063	NM_012249	Huesken	TC10	23433	CAUUAUCUGAUGCAUGGCAua	UGCCAUGCAUCAGAUAAUG	-2.1	-2.1	0.1	0	-1	3	-36.2	65.8	-0.7	5	II	42.1	45.8	59.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0064	NM_012249	Huesken	TC10	23433	UUAUCUGAUGCAUGGCAUAug	UAUGCCAUGCAUCAGAUAA	-1.3	-0.9	-0.7	-1	0	3	-35.4	70.9	-6.1	7	II	36.8	59.7	75.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0065	NM_012249	Huesken	TC10	23433	AUGCAAUAAGGCAGAUCCAca	UGGAUCUGCCUUAUUGCAU	-2.1	-1.1	-1.1	2	1	3	-37.2	51.8	-6	4	II	42.1	50.4	88.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0066	NM_012249	Huesken	TC10	23433	GCAAUAAGGCAGAUCCACAgc	UGUGGAUCUGCCUUAUUGC	-2.1	-3.4	-0.2	1	-1	3	-38.3	50.9	-1.7	4	II	47.4	39.8	68.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0067	NM_012249	Huesken	TC10	23433	CAAUAAGGCAGAUCCACAGca	CUGUGGAUCUGCCUUAUUG	-2.1	-2.1	-0.2	0	1	3	-37	61	0.7	6	II	47.4	52.8	79	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0068	NM_004359	Huesken	CDC34	997	CGUCCCGUAGUAGUCGUCGua	CGACGACUACUACGGGACG	-2.4	-2.4	0.6	0	1	4	-41.2	47.7	0.3	2	II	63.2	47	62.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0069	NM_004359	Huesken	CDC34	997	CCCGUAGUAGUCGUCGUAGaa	CUACGACGACUACUACGGG	-2.1	-3.3	0.6	-1	-1	4	-39.8	44.2	-2	2	II	57.9	38.2	69	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0070	NM_004359	Huesken	CDC34	997	GUAGUCGUCGUAGAAGAGGuc	CCUCUUCUACGACGACUAC	-3.3	-2.2	1.4	1	2	2	-38.4	46	0	4	II	52.6	53	82.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0071	NM_004359	Huesken	CDC34	997	UAGUCGUCGUAGAAGAGGUcu	ACCUCUUCUACGACGACUA	-2.2	-1.3	1.4	2	2	2	-38.4	67.7	-4.3	6	II	47.4	62.9	89.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0072	NM_004359	Huesken	CDC34	997	GUCGUAGAAGAGGUCUGAGcc	CUCAGACCUCUUCUACGAC	-2.1	-2.2	0.4	0	2	2	-39.1	50	4.6	3	II	52.6	39.7	37.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0073	NM_004359	Huesken	CDC34	997	AGAGGUCUGAGCCCUCGUCgg	GACGAGGGCUCAGACCUCU	-2.4	-2.1	-5.8	2	3	4	-44.7	48.9	7.5	2	II	63.2	53	51.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0074	NM_004359	Huesken	CDC34	997	GUCUGAGCCCUCGUCGGGCgc	GCCCGACGAGGGCUCAGAC	-3.4	-2.2	-7.8	1	3	5	-47.2	35.3	7.4	1	II	73.7	43.3	33.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0075	NM_004359	Huesken	CDC34	997	GUCGGGCGCCGGCGCCUUGgu	CAAGGCGCCGGCGCCCGAC	-2.1	-2.2	-6.9	0	0	14	-49.4	18.3	2.4	0	II	84.2	30.7	26.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0076	NM_004359	Huesken	CDC34	997	GCGCCGGCGCCUUGGUCUUca	AAGACCAAGGCGCCGGCGC	-0.9	-3.4	-2.6	1	-2	11	-46.7	25.6	-3.3	-1	III	73.7	21.7	25.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0077	NM_004359	Huesken	CDC34	997	UGGUCUUCACGCAGUACUCgg	GAGUACUGCGUGAAGACCA	-2.4	-2.1	0.3	0	2	3	-39.9	68.2	2.5	5	Ib	52.6	62.3	73.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0078	NM_004359	Huesken	CDC34	997	GUCUUCACGCAGUACUCGGcc	CCGAGUACUGCGUGAAGAC	-3.3	-2.2	0.3	1	2	3	-40.2	61.4	2.3	3	II	57.9	49.6	59	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0079	NM_004359	Huesken	CDC34	997	UUCACGCAGUACUCGGCCAgc	UGGCCGAGUACUGCGUGAA	-2.1	-0.9	-0.3	1	1	5	-42.3	58.9	-3.8	4	II	57.9	54.1	65.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0080	NM_004359	Huesken	CDC34	997	CGCAGUACUCGGCCAGCGUgg	ACGCUGGCCGAGUACUGCG	-2.2	-2.4	-1.6	-1	-2	5	-44.8	37.9	-3.6	0	III	68.4	34.3	52.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0081	NM_004359	Huesken	CDC34	997	GUCACGCUCCGCGUCCACCuu	GGUGGACGCGGAGCGUGAC	-3.3	-2.2	-2.9	1	2	5	-46.2	45.4	7.4	0	II	73.7	46.3	53.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0082	NM_004359	Huesken	CDC34	997	CCUUGGUCCCCAGGACCUGcu	CAGGUCCUGGGGACCAAGG	-2.1	-3.3	-7.4	0	0	4	-45.8	28.9	-4.4	1	II	68.4	37.3	43	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0083	NM_004359	Huesken	CDC34	997	GUCCCCAGGACCUGCUUCCgg	GGAAGCAGGUCCUGGGGAC	-3.3	-2.2	-2.2	1	1	4	-46.2	48.2	2.7	2	II	68.4	47.5	58.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0084	NM_004359	Huesken	CDC34	997	GCUUCCGGAUGAUGUCUGUgu	ACAGACAUCAUCCGGAAGC	-2.2	-3.4	1.2	-1	-1	4	-40	52.9	-4	4	III	52.6	27.2	34.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0085	NM_004359	Huesken	CDC34	997	GAUGAUGUCUGUGUACUCCcg	GGAGUACACAGACAUCAUC	-3.3	-2.4	0.7	1	3	2	-37.8	69.5	15.5	5	II	47.4	55.1	76.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0086	NM_004359	Huesken	CDC34	997	UGUGUACUCCCGAUCCUUCcc	GAAGGAUCGGGAGUACACA	-2.4	-2.1	1.2	1	3	4	-40.2	66.1	4.8	6	Ib	52.6	62.6	74.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0087	NM_004359	Huesken	CDC34	997	GUGUACUCCCGAUCCUUCCcc	GGAAGGAUCGGGAGUACAC	-3.3	-2.2	1.2	2	1	4	-41.4	67.7	10.1	4	II	57.9	55.5	56.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0088	NM_004359	Huesken	CDC34	997	CCCGAUCCUUCCCCUUGCUcu	AGCAAGGGGAAGGAUCGGG	-2.1	-3.3	-0.9	-1	-2	4	-44.1	39.7	-5.6	0	III	63.2	30.7	41.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0089	NM_004359	Huesken	CDC34	997	UUUCCACUUCCUGUACAUCac	GAUGUACAGGAAGUGGAAA	-2.4	-0.9	0.4	1	4	2	-36	75.1	10.4	5	Ib	42.1	64.9	82.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0090	NM_004359	Huesken	CDC34	997	CCUGUACAUCACGGAGGCGuc	CGCCUCCGUGAUGUACAGG	-2.4	-3.3	-0.3	-2	1	4	-42.4	31	-2.1	1	II	63.2	35.3	36.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0091	NM_004359	Huesken	CDC34	997	CAUCACGGAGGCGUCCACGuu	CGUGGACGCCUCCGUGAUG	-2.4	-2.1	-2.9	0	2	4	-43.8	41.4	0.7	2	II	68.4	49.6	79.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0092	NM_004359	Huesken	CDC34	997	UCACGGAGGCGUCCACGUUug	AACGUGGACGCCUCCGUGA	-0.9	-2.4	-3.4	2	1	4	-43.7	38.9	1.5	3	II	63.2	48.6	55.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0093	NM_004359	Huesken	CDC34	997	CGGGCGAGAAGGUGUUGGGcu	CCCAACACCUUCUCGCCCG	-3.3	-2.4	4.6	-2	0	6	-44.1	37.9	-2	1	II	68.4	37.9	45.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0094	NM_004359	Huesken	CDC34	997	GUUGGGCUCGUUCAGGAGGga	CCUCCUGAACGAGCCCAAC	-3.3	-2.2	-0.7	1	2	4	-42.9	45	5.4	1	II	63.2	46.9	68.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0095	NM_004359	Huesken	CDC34	997	UCGUUCAGGAGGGAGAUCAca	UGAUCUCCCUCCUGAACGA	-2.1	-2.4	0.6	1	-1	3	-41.4	59.7	-3.8	8	II	52.6	56.7	70	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0096	NM_004359	Huesken	CDC34	997	GGAGGGAGAUCACACUCAGga	CUGAGUGUGAUCUCCCUCC	-2.1	-3.3	-1.4	0	1	3	-42.1	33.2	0.1	0	II	57.9	37.1	23.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0097	NM_004359	Huesken	CDC34	997	GAGGGAGAUCACACUCAGGag	CCUGAGUGUGAUCUCCCUC	-3.3	-2.4	-1.4	0	1	3	-42.1	40.2	-2.6	0	II	57.9	44.9	58.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0098	NM_004359	Huesken	CDC34	997	GGGAGAUCACACUCAGGAGaa	CUCCUGAGUGUGAUCUCCC	-2.1	-3.3	-1.4	1	0	3	-42.1	48.2	-0.1	0	II	57.9	34.7	42.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0099	NM_004359	Huesken	CDC34	997	UCACACUCAGGAGAAUGGUcc	ACCAUUCUCCUGAGUGUGA	-2.2	-2.4	-3.8	2	2	2	-39.5	52.6	-1.3	5	II	47.4	55.2	81.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0100	NM_004359	Huesken	CDC34	997	CACACUCAGGAGAAUGGUCcu	GACCAUUCUCCUGAGUGUG	-2.4	-2.1	-3.8	0	0	2	-39.5	55.8	5.1	1	II	52.6	47.3	44.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0101	NM_004359	Huesken	CDC34	997	CAGGAGAAUGGUCCUGACGuu	CGUCAGGACCAUUCUCCUG	-2.4	-2.1	-4.8	-1	0	2	-40.9	51.1	6.1	1	II	57.9	53.3	72.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0102	NM_004359	Huesken	CDC34	997	AGGAGAAUGGUCCUGACGUuc	ACGUCAGGACCAUUCUCCU	-2.2	-2.1	-2.4	2	2	2	-41	52.4	-1.6	3	II	52.6	51.6	72.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0103	NM_004359	Huesken	CDC34	997	CUGACGUUCUGCGUGGGGUuc	ACCCCACGCAGAACGUCAG	-2.2	-2.1	0	1	0	4	-43.1	52.4	-6	2	III	63.2	43	57.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0104	NM_004359	Huesken	CDC34	997	UCUGCGUGGGGUUCCACCUcu	AGGUGGAACCCCACGCAGA	-2.1	-2.4	-5.2	2	2	4	-45.1	58.4	-1	3	II	63.2	60	72.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0105	NM_004359	Huesken	CDC34	997	CCACCUCUCUGAGGGCAGCuc	GCUGCCCUCAGAGAGGUGG	-3.4	-3.3	-3.2	0	0	4	-46.2	33.2	5.4	-1	II	68.4	36.1	39.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0106	NM_004359	Huesken	CDC34	997	GGUCGUCCACCGGCGGGUGga	CACCCGCCGGUGGACGACC	-2.1	-3.3	-4.2	1	1	8	-48.2	23.4	-2.4	-1	II	78.9	27.8	25.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0107	NM_004359	Huesken	CDC34	997	GUCGUCCACCGGCGGGUGGag	CCACCCGCCGGUGGACGAC	-3.3	-2.2	-7.3	0	1	8	-48.2	32.5	2.4	1	II	78.9	33.5	44.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0108	NM_004359	Huesken	CDC34	997	CGUAGAUGUUAGGGUGCCAca	UGGCACCCUAACAUCUACG	-2.1	-2.4	1.2	-1	-2	3	-39.8	42.1	-3.5	3	II	52.6	31.1	25.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0109	NM_004359	Huesken	CDC34	997	UGUUAGGGUGCCACAUCUUgg	AAGAUGUGGCACCCUAACA	-0.9	-2.1	0.9	0	1	3	-39.1	56.3	-8.6	7	II	47.4	55.3	36.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0110	NM_004359	Huesken	CDC34	997	GUGGAGAGUAUGGGUAGUCga	GACUACCCAUACUCUCCAC	-2.4	-2.2	2.7	-1	1	3	-40.1	40.6	10.1	4	II	52.6	48.3	65.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0111	NM_004359	Huesken	CDC34	997	GUAUGGGUAGUCGAUGGGGaa	CCCCAUCGACUACCCAUAC	-3.3	-2.2	4.1	0	3	4	-41.4	42.3	-2.4	3	II	57.9	49.2	44.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0112	NM_004359	Huesken	CDC34	997	GAUGGGGAACUUGAGGCGCgc	GCGCCUCAAGUUCCCCAUC	-3.4	-2.4	1.6	0	3	5	-43.1	36.5	10.4	2	II	63.2	47.2	58	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0113	NM_004359	Huesken	CDC34	997	AUGGGGAACUUGAGGCGCGcc	CGCGCCUCAAGUUCCCCAU	-2.4	-1.1	1.3	1	4	6	-43.1	34.7	-2.6	3	II	63.2	59.3	51.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0114	NM_004359	Huesken	CDC34	997	CGCCUUGAAGUAGCCGCCCuc	GGGCGGCUACUUCAAGGCG	-3.3	-2.4	-1.8	0	0	7	-44.3	34.4	5.4	0	II	68.4	47.1	29.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0115	NM_004359	Huesken	CDC34	997	GAAGUAGCCGCCCUCGUAGua	CUACGAGGGCGGCUACUUC	-2.1	-2.4	-1.9	1	2	7	-42.6	47.6	0	3	II	63.2	45.1	59.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0116	NM_004359	Huesken	CDC34	997	AAGUAGCCGCCCUCGUAGUag	ACUACGAGGGCGGCUACUU	-2.2	-0.9	0	2	1	7	-42.4	52.7	-6	4	II	57.9	47.2	73.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0117	NM_004359	Huesken	CDC34	997	CCUCCCAGUUGUAUAGAUCgc	GAUCUAUACAACUGGGAGG	-2.4	-3.3	-0.1	0	0	3	-37.7	62.4	8.2	2	II	47.4	44.2	41.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0118	NM_004359	Huesken	CDC34	997	UCCCAGUUGUAUAGAUCGCcc	GCGAUCUAUACAACUGGGA	-3.4	-2.4	-0.1	2	3	3	-38.1	60.7	12.8	5	II	47.4	59.4	74	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0119	NM_004359	Huesken	CDC34	997	GUUGUAUAGAUCGCCCUCGuc	CGAGGGCGAUCUAUACAAC	-2.4	-2.2	-0.1	2	2	5	-38.4	58.2	-2.4	6	II	52.6	58.7	72.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0120	NM_004359	Huesken	CDC34	997	UGUAUAGAUCGCCCUCGUCca	GACGAGGGCGAUCUAUACA	-2.4	-2.1	-0.1	0	3	5	-39.9	64.6	7.5	7	Ia	52.6	61.2	64.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0121	NM_004359	Huesken	CDC34	997	UCGCCCUCGUCCACCAGUGuc	CACUGGUGGACGAGGGCGA	-2.1	-2.4	0.3	3	2	5	-45.7	46.1	-7.3	3	II	68.4	56.7	75.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0122	NM_004359	Huesken	CDC34	997	CGUCCACCAGUGUCACGCGga	CGCGUGACACUGGUGGACG	-2.4	-2.4	-0.4	0	1	4	-43.5	37.9	-2	0	II	68.4	38.9	67.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0123	NM_004359	Huesken	CDC34	997	CCCUCGACCGGCUCUUCCUgc	AGGAAGAGCCGGUCGAGGG	-2.1	-3.3	-1.6	0	-2	5	-45.8	48	-8.3	0	III	68.4	37.7	44.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0124	NM_004359	Huesken	CDC34	997	CGGCUCUUCCUGCAGCCCCuu	GGGGCUGCAGGAAGAGCCG	-3.3	-2.4	-5.6	2	0	5	-47.4	41.9	7.8	0	II	73.7	49.5	47.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0125	NM_003340	Huesken	UBE2D3	7323	GUCUGUUUUAUAGAUCCGUgc	ACGGAUCUAUAAAACAGAC	-2.2	-2.2	0.8	1	0	3	-33.3	68.4	-4	4	II	36.8	51.4	68.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0126	NM_003340	Huesken	UBE2D3	7323	UUUAUAGAUCCGUGCAAUCuc	GAUUGCACGGAUCUAUAAA	-2.4	-0.9	-0.7	-1	3	3	-33.4	77.6	5.1	8	Ia	36.8	71.1	92.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0127	NM_003340	Huesken	UBE2D3	7323	AGAUCCGUGCAAUCUCUGGca	CCAGAGAUUGCACGGAUCU	-3.3	-2.1	-1.5	2	3	3	-39.9	67.4	2.3	5	II	52.6	56.3	82	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0128	NM_003340	Huesken	UBE2D3	7323	CGUGCAAUCUCUGGCACUAgg	UAGUGCCAGAGAUUGCACG	-1.3	-2.4	-4.8	-1	-3	3	-39.7	43.5	-5.1	1	III	52.6	29.4	62.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0129	NM_003340	Huesken	UBE2D3	7323	UGCAAUCUCUGGCACUAGGgg	CCUAGUGCCAGAGAUUGCA	-3.3	-2.1	-0.4	1	2	3	-40.5	66.3	5.1	5	Ib	52.6	59.2	86.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0130	NM_003340	Huesken	UBE2D3	7323	GGGUCAUCUGGGUUUGGAUca	AUCCAAACCCAGAUGACCC	-1.1	-3.3	0.9	1	-1	3	-40.5	60.4	-1.6	2	III	52.6	28.8	44.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0131	NM_003340	Huesken	UBE2D3	7323	GGUCAUCUGGGUUUGGAUCac	GAUCCAAACCCAGAUGACC	-2.4	-3.3	0.5	2	1	3	-39.6	64.6	15.5	2	II	52.6	41.2	56.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0132	NM_003340	Huesken	UBE2D3	7323	CUGGGUUUGGAUCACAUAGca	CUAUGUGAUCCAAACCCAG	-2.1	-2.1	1.5	1	-1	3	-37	67.6	5.7	3	II	47.4	54.8	95.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0133	NM_003340	Huesken	UBE2D3	7323	GGGUUUGGAUCACAUAGCAgu	UGCUAUGUGAUCCAAACCC	-2.1	-3.3	1.5	0	-2	3	-38.3	51.7	-3.7	3	III	47.4	33.6	42.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0134	NM_003340	Huesken	UBE2D3	7323	UUGGAUCACAUAGCAGUGAac	UCACUGCUAUGUGAUCCAA	-2.4	-0.9	-3.4	1	0	2	-37.4	59.8	-1.1	6	II	42.1	60.8	95.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0135	NM_003340	Huesken	UBE2D3	7323	CAUAGCAGUGAACAAAUGGau	CCAUUUGUUCACUGCUAUG	-3.3	-2.1	1.1	-1	1	2	-34.4	65	2.7	5	II	42.1	57.5	94.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0136	NM_003340	Huesken	UBE2D3	7323	AUAGCAGUGAACAAAUGGAua	UCCAUUUGUUCACUGCUAU	-2.4	-1.1	1.1	2	2	2	-34.7	61.8	-6.4	6	II	36.8	56.1	98.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0137	NM_003340	Huesken	UBE2D3	7323	UAGCAGUGAACAAAUGGAUaa	AUCCAUUUGUUCACUGCUA	-1.1	-1.3	1.1	2	1	2	-34.7	63.7	-6.2	7	II	36.8	61.1	96.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0138	NM_003340	Huesken	UBE2D3	7323	GCAGUGAACAAAUGGAUAAaa	UUAUCCAUUUGUUCACUGC	-0.9	-3.4	2.3	0	-3	2	-33.5	57.5	-1.5	5	III	36.8	35.5	23.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0139	NM_003340	Huesken	UBE2D3	7323	GUGAACAAAUGGAUAAAAGaa	CUUUUAUCCAUUUGUUCAC	-2.1	-2.2	3	0	1	2	-29.8	71.2	2.4	4	II	31.6	48.7	27	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0140	NM_003340	Huesken	UBE2D3	7323	UGGAUAAAAGAACUUUAGAaa	UCUAAAGUUCUUUUAUCCA	-2.4	-2.1	2.3	0	0	2	-30.2	81.8	3.3	6	II	26.3	57.8	64.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0141	NM_003340	Huesken	UBE2D3	7323	GAUAAAAGAACUUUAGAAAuu	UUUCUAAAGUUCUUUUAUC	-0.9	-2.4	2.3	0	-1	1	-26.6	83	2	6	II	21.1	41.6	30.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0142	NM_003340	Huesken	UBE2D3	7323	AAAAGAACUUUAGAAAUUGuu	CAAUUUCUAAAGUUCUUUU	-2.1	-0.9	2.5	1	3	1	-25.9	73	0	7	Ia	21.1	68.5	80.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0143	NM_003340	Huesken	UBE2D3	7323	UUUAGAAAUUGUUAAAGCAgg	UGCUUUAACAAUUUCUAAA	-2.1	-0.9	1.9	0	2	2	-27.3	88.2	-0.7	7	II	21.1	67.6	92	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0144	NM_003340	Huesken	UBE2D3	7323	UUAGAAAUUGUUAAAGCAGgc	CUGCUUUAACAAUUUCUAA	-2.1	-0.9	1.8	0	4	2	-28.5	76.1	3.1	7	Ia	26.3	75.8	89.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0145	NM_003340	Huesken	UBE2D3	7323	AUUGUUAAAGCAGGCGACCac	GGUCGCCUGCUUUAACAAU	-3.3	-1.1	1.2	0	4	4	-37	59.1	5.1	6	Ia	47.4	67.5	84.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0146	NM_003340	Huesken	UBE2D3	7323	UUGUUAAAGCAGGCGACCAcu	UGGUCGCCUGCUUUAACAA	-2.1	-0.9	1.2	1	1	4	-38	63	-1.5	7	II	47.4	61.5	90.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0147	NM_003340	Huesken	UBE2D3	7323	UGUUAAAGCAGGCGACCACug	GUGGUCGCCUGCUUUAACA	-2.2	-2.1	-0.9	0	3	4	-39.3	58.6	12.5	8	Ia	52.6	57.2	100.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0148	NM_003340	Huesken	UBE2D3	7323	GGCGACCACUGUGAUCUUAga	UAAGAUCACAGUGGUCGCC	-1.3	-3.3	-3.4	1	-4	4	-40	42.6	2	1	III	52.6	21.3	55.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0149	NM_003340	Huesken	UBE2D3	7323	CGACCACUGUGAUCUUAGAau	UCUAAGAUCACAGUGGUCG	-2.4	-2.4	-3.4	1	-3	2	-37.8	51.6	-5.8	0	III	47.4	27.4	43.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0150	NM_003340	Huesken	UBE2D3	7323	UGAUCUUAGAAUAUCGAGAca	UCUCGAUAUUCUAAGAUCA	-2.4	-2.1	-0.5	2	1	2	-32.9	90.6	1.3	7	II	31.6	66.1	98.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0151	NM_003340	Huesken	UBE2D3	7323	GAAUAUCGAGACAAAUGCUgc	AGCAUUUGUCUCGAUAUUC	-2.1	-2.4	1.4	0	1	2	-33.3	62.7	-3.9	4	II	36.8	46.6	79.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0152	NM_003340	Huesken	UBE2D3	7323	AAUAUCGAGACAAAUGCUGcc	CAGCAUUUGUCUCGAUAUU	-2.1	-0.9	1.4	2	4	2	-33	71.4	-4.9	8	Ia	36.8	67.7	95.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0153	NM_003340	Huesken	UBE2D3	7323	AUCGAGACAAAUGCUGCCAuu	UGGCAGCAUUUGUCUCGAU	-2.1	-1.1	-1.7	1	1	3	-38.5	59.8	4.3	5	II	47.4	55.7	99.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0154	NM_003340	Huesken	UBE2D3	7323	ACAAAUGCUGCCAUUACUGuu	CAGUAAUGGCAGCAUUUGU	-2.1	-2.2	-0.4	1	4	3	-35.4	65.3	-2.3	6	Ia	42.1	63.7	97.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0155	NM_003340	Huesken	UBE2D3	7323	UGCUGCCAUUACUGUUAAUau	AUUAACAGUAAUGGCAGCA	-1.1	-2.1	-0.9	-1	-1	3	-34.6	70.9	-4	4	II	36.8	41.1	57.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0156	NM_003340	Huesken	UBE2D3	7323	CUGUUAAUAUUUGGAUGAUaa	AUCAUCCAAAUAUUAACAG	-1.1	-2.1	3.8	-2	-2	2	-29.3	69.7	-3	4	II	26.3	37.9	32	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0157	NM_003340	Huesken	UBE2D3	7323	UAAUAUUUGGAUGAUAAAUuc	AUUUAUCAUCCAAAUAUUA	-1.1	-1.3	2.5	2	1	2	-26.2	99.4	4.1	8	II	15.8	70.4	76.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0158	NM_003340	Huesken	UBE2D3	7323	UUUGGAUGAUAAAUUCUUGuu	CAAGAAUUUAUCAUCCAAA	-2.1	-0.9	1	1	4	2	-28.9	86.7	2	7	Ia	26.3	73.2	93.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0159	NM_003340	Huesken	UBE2D3	7323	GAUAAAUUCUUGUUGUAAAug	UUUACAACAAGAAUUUAUC	-0.9	-2.4	2.7	2	-1	1	-26.6	87.6	1.3	5	II	21.1	41.6	48.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0160	NM_003340	Huesken	UBE2D3	7323	AUAAAUUCUUGUUGUAAAUgc	AUUUACAACAAGAAUUUAU	-1.1	-1.1	2.4	1	1	1	-25.3	96.6	1.8	7	II	15.8	59.7	74.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0161	NM_003340	Huesken	UBE2D3	7323	UAAAUUCUUGUUGUAAAUGca	CAUUUACAACAAGAAUUUA	-2.1	-1.3	2.2	0	4	1	-26.3	97	5.4	7	Ia	21.1	78.7	70	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0162	NM_003340	Huesken	UBE2D3	7323	AAUUCUUGUUGUAAAUGCAac	UGCAUUUACAACAAGAAUU	-2.1	-0.9	1.9	2	2	2	-29.6	88.6	-3.3	6	II	26.3	58.9	83.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0163	NM_003340	Huesken	UBE2D3	7323	AUUCUUGUUGUAAAUGCAAcc	UUGCAUUUACAACAAGAAU	-0.9	-1.1	1.3	2	1	2	-29.6	73.3	-6.3	6	II	26.3	57	68.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0164	NM_003340	Huesken	UBE2D3	7323	CUUGUUGUAAAUGCAACCUua	AGGUUGCAUUUACAACAAG	-2.1	-2.1	-2.6	-1	0	2	-32.8	76.3	-0.7	4	II	36.8	57.3	86.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0165	NM_003340	Huesken	UBE2D3	7323	UUGUUGUAAAUGCAACCUUag	AAGGUUGCAUUUACAACAA	-0.9	-0.9	-2.6	1	1	2	-31.6	79.9	-1.6	7	II	31.6	70.7	83	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0166	NM_003340	Huesken	UBE2D3	7323	GUUGUAAAUGCAACCUUAGgu	CUAAGGUUGCAUUUACAAC	-2.1	-2.2	-0.9	0	1	2	-32	65.2	-4.9	5	II	36.8	51.4	80.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0167	NM_003340	Huesken	UBE2D3	7323	CAACCUUAGGUGGUUUGAAgg	UUCAAACCACCUAAGGUUG	-0.9	-2.1	-3	2	-1	2	-35.2	59.3	-3.4	2	II	42.1	39.2	64.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0168	NM_003340	Huesken	UBE2D3	7323	ACCUUAGGUGGUUUGAAGGgg	CCUUCAAACCACCUAAGGU	-3.3	-2.2	-1.4	1	3	2	-37.6	71.5	5.4	4	Ib	47.4	54.8	94.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0169	NM_003340	Huesken	UBE2D3	7323	GGUGGUUUGAAGGGGUAGUcu	ACUACCCCUUCAAACCACC	-2.2	-3.3		1	-1	4	-40.1	50.1	-1.7	3	III	52.6	35.6	63.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0170	NM_003340	Huesken	UBE2D3	7323	GUGGUUUGAAGGGGUAGUCug	GACUACCCCUUCAAACCAC	-2.4	-2.2	1.3	0	1	4	-39.2	55.3	7.4	4	II	52.6	48.6	87.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0171	NM_003340	Huesken	UBE2D3	7323	UUGAAGGGGUAGUCUGUAGga	CUACAGACUACCCCUUCAA	-2.1	-0.9	1.1	1	2	4	-38.3	68.4	5	8	Ib	47.4	62.9	94.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0172	NM_003340	Huesken	UBE2D3	7323	UGAAGGGGUAGUCUGUAGGaa	CCUACAGACUACCCCUUCA	-3.3	-2.1	1.1	0	4	4	-40.7	67.4	7.8	7	II	52.6	67	99.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0173	NM_003340	Huesken	UBE2D3	7323	UAGUCUGUAGGAAAAUGAAuu	UUCAUUUUCCUACAGACUA	-0.9	-1.3	2	0	0	2	-32.7	75.2	-6.3	6	II	31.6	55.7	85	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0174	NM_003340	Huesken	UBE2D3	7323	GUCUGUAGGAAAAUGAAUUgu	AAUUCAUUUUCCUACAGAC	-0.9	-2.2	2	1	-1	2	-31.3	74.3	-1.9	4	II	31.6	49.8	62.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0175	NM_003340	Huesken	UBE2D3	7323	CUGUAGGAAAAUGAAUUGUca	ACAAUUCAUUUUCCUACAG	-2.2	-2.1	3.5	-1	-2	2	-31	70.2	2.1	6	III	31.6	49.9	76.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0176	NM_003340	Huesken	UBE2D3	7323	UGUAGGAAAAUGAAUUGUCaa	GACAAUUCAUUUUCCUACA	-2.4	-2.1	1.3	0	3	2	-31.3	75.6	4.8	7	Ib	31.6	72.2	92.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0177	NM_003340	Huesken	UBE2D3	7323	GUAGGAAAAUGAAUUGUCAaa	UGACAAUUCAUUUUCCUAC	-2.1	-2.2	0.8	0	0	2	-31.3	62.6	-4	4	II	31.6	44.9	55.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0178	NM_003340	Huesken	UBE2D3	7323	AAAAGAAUACACCGCCUUGau	CAAGGCGGUGUAUUCUUUU	-2.1	-0.9	1	0	3	5	-34.5	65.6	2.4	7	Ia	42.1	61.3	104.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0179	NM_003340	Huesken	UBE2D3	7323	AGAAUACACCGCCUUGAUAug	UAUCAAGGCGGUGUAUUCU	-1.3	-2.1	0.1	2	0	5	-36.6	63.6	-1.4	5	II	42.1	42.2	99.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0180	NM_003340	Huesken	UBE2D3	7323	CUUGAUAUGGGCUGUCAUUag	AAUGACAGCCCAUAUCAAG	-0.9	-2.1	0.2	-1	-2	4	-36	63.2	-5.6	4	II	42.1	45.8	80	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0181	NM_003340	Huesken	UBE2D3	7323	UUGAUAUGGGCUGUCAUUAgg	UAAUGACAGCCCAUAUCAA	-1.3	-0.9	0.2	2	0	4	-35.2	78.2	-0.8	8	II	36.8	63.3	97.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0182	NM_003340	Huesken	UBE2D3	7323	GGGCUGUCAUUAGGUCCCAua	UGGGACCUAAUGACAGCCC	-2.1	-3.3	-1.2	1	-2	4	-42.9	31.6	-3.8	0	III	57.9	29.3	48.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0183	NM_003340	Huesken	UBE2D3	7323	GCUGUCAUUAGGUCCCAUAau	UAUGGGACCUAAUGACAGC	-1.3	-3.4	1.5	0	-3	3	-38.7	61.6	-3.8	4	III	47.4	36.9	44.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0184	NM_003340	Huesken	UBE2D3	7323	UGUCAUUAGGUCCCAUAAUug	AUUAUGGGACCUAAUGACA	-1.1	-2.1	0.5	3	0	3	-35.2	70.2	-1.2	6	II	36.8	56	64.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0185	NM_003340	Huesken	UBE2D3	7323	GGUCCCAUAAUUGUGGCUUgc	AAGCCACAAUUAUGGGACC	-0.9	-3.3	-1.6	1	0	3	-38	62.9	1.4	2	III	47.4	38.6	52	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0186	NM_003340	Huesken	UBE2D3	7323	GUCCCAUAAUUGUGGCUUGcc	CAAGCCACAAUUAUGGGAC	-2.1	-2.2	-1.6	1	0	3	-36.8	57.7	0	2	II	47.4	41.1	35.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0187	NM_003340	Huesken	UBE2D3	7323	UCCCAUAAUUGUGGCUUGCca	GCAAGCCACAAUUAUGGGA	-3.4	-2.4	-1.6	0	2	3	-38	63.1	13.1	5	Ib	47.4	59.2	75.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0188	NM_003340	Huesken	UBE2D3	7323	CCCAUAAUUGUGGCUUGCCaa	GGCAAGCCACAAUUAUGGG	-3.3	-3.3	-1.8	-1	0	3	-38.9	53.8	5.4	2	II	52.6	48.6	60.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0189	NM_003340	Huesken	UBE2D3	7323	UAAUUGUGGCUUGCCAAUGaa	CAUUGGCAAGCCACAAUUA	-2.1	-1.3	-2.4	2	3	3	-35.3	75.5	3	7	Ia	42.1	74.6	100.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0190	NM_003340	Huesken	UBE2D3	7323	UUGUGGCUUGCCAAUGAAAca	UUUCAUUGGCAAGCCACAA	-0.9	-0.9	-2.4	0	-1	3	-36.2	60.8	-11	4	II	42.1	52.6	85.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0191	NM_003340	Huesken	UBE2D3	7323	UGUGGCUUGCCAAUGAAACau	GUUUCAUUGGCAAGCCACA	-2.2	-2.1	-2.4	2	3	3	-37.5	67.8	4.8	4	II	47.4	63.5	94.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0192	NM_003340	Huesken	UBE2D3	7323	UGGCUUGCCAAUGAAACAUau	AUGUUUCAUUGGCAAGCCA	-1.1	-2.1	-1.9	2	0	3	-36.4	67.9	1.4	5	II	42.1	58	90.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0193	NM_003340	Huesken	UBE2D3	7323	GGCUUGCCAAUGAAACAUAuc	UAUGUUUCAUUGGCAAGCC	-1.3	-3.3	-0.9	0	-4	3	-35.6	50.8	-6.3	3	III	42.1	31.1	44.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0194	NM_003340	Huesken	UBE2D3	7323	GCCAAUGAAACAUAUCAUCcc	GAUGAUAUGUUUCAUUGGC	-2.4	-3.4	-2.8	0	-1	3	-32.9	65.4	7.8	3	II	36.8	48.3	66.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0195	NM_003340	Huesken	UBE2D3	7323	CAAUGAAACAUAUCAUCCCca	GGGAUGAUAUGUUUCAUUG	-3.3	-2.1	-3.2	0	2	3	-32.8	70.9	5.4	4	II	36.8	65	95	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0196	NM_003340	Huesken	UBE2D3	7323	AUAUCAUCCCCAACUGGACcu	GUCCAGUUGGGGAUGAUAU	-2.2	-1.1	-3.3	3	4	4	-38.7	67	2.5	5	Ia	47.4	60.7	88.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0197	NM_003340	Huesken	UBE2D3	7323	UAUCAUCCCCAACUGGACCug	GGUCCAGUUGGGGAUGAUA	-3.3	-1.3	-3.7	2	5	4	-40.9	72.5	17.1	5	Ia	52.6	71.8	93.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0198	NM_003340	Huesken	UBE2D3	7323	CCCCAACUGGACCUGCAGAac	UCUGCAGGUCCAGUUGGGG	-2.4	-3.3	-1.5	1	-3	4	-44.7	35.1	1.6	-1	III	63.2	24.7	32.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0199	NM_003340	Huesken	UBE2D3	7323	CCCAACUGGACCUGCAGAAca	UUCUGCAGGUCCAGUUGGG	-0.9	-3.3	-0.9	0	-4	3	-42.3	41.8	-7.8	3	III	57.9	27.3	50	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0200	NM_003340	Huesken	UBE2D3	7323	GACCUGCAGAACAUUGUGCug	GCACAAUGUUCUGCAGGUC	-3.4	-2.4	-1.8	2	2	2	-39.1	59.8	7.1	3	II	52.6	49	100.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0201	NM_003340	Huesken	UBE2D3	7323	ACCUGCAGAACAUUGUGCUgg	AGCACAAUGUUCUGCAGGU	-2.1	-2.2	-2	1	2	2	-38.8	65.1	-4	4	II	47.4	50.6	101.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0202	NM_003340	Huesken	UBE2D3	7323	CCUGCAGAACAUUGUGCUGga	CAGCACAAUGUUCUGCAGG	-2.1	-3.3	-2	-2	0	2	-38.7	51.2	3.3	1	II	52.6	36.8	87.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0203	NM_003337	Huesken	UBE2B	7320	CUUUGUUCAACAAUGGCCGaa	CGGCCAUUGUUGAACAAAG	-2.4	-2.1	-0.7	0	3	5	-35.3	73.2	-4.6	5	II	47.4	64.5	76.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0204	NM_003337	Huesken	UBE2B	7320	GUUCAACAAUGGCCGAAACuc	GUUUCGGCCAUUGUUGAAC	-2.2	-2.2	-3.1	1	1	5	-35.7	55.5	12.5	2	II	47.4	37.9	61.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0205	NM_003337	Huesken	UBE2B	7320	CAAUGGCCGAAACUCUUUUcu	AAAAGAGUUUCGGCCAUUG	-0.9	-2.1	1.5	0	-1	5	-34.3	66.2	1.1	5	III	42.1	46.5	49.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0206	NM_003337	Huesken	UBE2B	7320	AUGGCCGAAACUCUUUUCUca	AGAAAAGAGUUUCGGCCAU	-2.1	-1.1	-0.8	1	2	5	-35.8	78.2	1.4	6	II	42.1	59.6	77.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0207	NM_003337	Huesken	UBE2B	7320	UGGCCGAAACUCUUUUCUCau	GAGAAAAGAGUUUCGGCCA	-2.4	-2.1	-0.9	1	3	5	-37.1	69.2	7.4	4	II	47.4	64.5	82.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0208	NM_003337	Huesken	UBE2B	7320	CGAAACUCUUUUCUCAUAUuc	AUAUGAGAAAAGAGUUUCG	-1.1	-2.4	-0.9	0	-2	2	-30.6	79.4	4.5	5	II	31.6	45.4	46.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0209	NM_003337	Huesken	UBE2B	7320	AAACUCUUUUCUCAUAUUCuc	GAAUAUGAGAAAAGAGUUU	-2.4	-0.9	3	3	3	1	-29.1	94	10.5	7	Ia	26.3	70.5	78.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0210	NM_003337	Huesken	UBE2B	7320	ACUCUUUUCUCAUAUUCUCgu	GAGAAUAUGAGAAAAGAGU	-2.4	-2.2	4.5	4	3	1	-31.8	85.3	5.1	5	Ia	31.6	63.6	83.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0211	NM_003337	Huesken	UBE2B	7320	CUCUUUUCUCAUAUUCUCGuu	CGAGAAUAUGAGAAAAGAG	-2.4	-2.1	3.1	0	1	2	-32	93.5	3.1	5	II	36.8	64.4	73.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0212	NM_003337	Huesken	UBE2B	7320	UCUUUUCUCAUAUUCUCGUuu	ACGAGAAUAUGAGAAAAGA	-2.2	-2.4	2.6	1	3	2	-32.1	93.7	-1.6	7	II	31.6	67.2	84.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0213	NM_003337	Huesken	UBE2B	7320	AUUCUCGUUUGUUUUCCUGau	CAGGAAAACAAACGAGAAU	-2.1	-1.1		2	5	2	-32.2	88.7	5.4	6	Ib	36.8	64.2	84	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0214	NM_003337	Huesken	UBE2B	7320	UUUGUUUUCCUGAUAAAGCug	GCUUUAUCAGGAAAACAAA	-3.4	-0.9	1.4	2	5	2	-30.8	94.1	7.1	7	Ia	31.6	84.8	80.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0215	NM_003337	Huesken	UBE2B	7320	UUUCCUGAUAAAGCUGUGCug	GCACAGCUUUAUCAGGAAA	-3.4	-0.9	1.1	1	4	2	-35.7	79.5	9.7	8	Ib	42.1	75.9	83.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0216	NM_003337	Huesken	UBE2B	7320	UCCUGAUAAAGCUGUGCUGcc	CAGCACAGCUUUAUCAGGA	-2.1	-2.4	0.3	0	2	2	-38.1	66.5	-2.4	6	Ib	47.4	57.3	77.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0217	NM_003337	Huesken	UBE2B	7320	GAUAAAGCUGUGCUGCCUGgc	CAGGCAGCACAGCUUUAUC	-2.1	-2.4	-2.1	0	3	3	-39.1	54.8	7.4	4	II	52.6	49.5	75.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0218	NM_003337	Huesken	UBE2B	7320	AAAGCUGUGCUGCCUGGCUau	AGCCAGGCAGCACAGCUUU	-2.1	-0.9	-2.1	2	3	3	-43.1	53.3	-1.6	3	II	57.9	55.9	71.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0219	NM_003337	Huesken	UBE2B	7320	AGCUGUGCUGCCUGGCUAUug	AUAGCCAGGCAGCACAGCU	-1.1	-2.1	-1.4	0	-1	3	-43.7	49.5	-8.7	3	II	57.9	38.3	71.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0220	NM_003337	Huesken	UBE2B	7320	GCUGUGCUGCCUGGCUAUUgg	AAUAGCCAGGCAGCACAGC	-0.9	-3.4	-0.9	0	-2	3	-42.5	37.1	-3.3	1	III	57.9	24.4	51	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0221	NM_003337	Huesken	UBE2B	7320	UGCUGCCUGGCUAUUGGCUgg	AGCCAAUAGCCAGGCAGCA	-2.1	-2.1	-4.5	1	2	3	-43.6	59.9	-3.6	2	II	57.9	50.8	75.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0222	NM_003337	Huesken	UBE2B	7320	CCUGGCUAUUGGCUGGACUgu	AGUCCAGCCAAUAGCCAGG	-2.1	-3.3	0.2	0	-1	3	-42.6	50.8	-1.2	1	III	57.9	38.1	68.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0223	NM_003337	Huesken	UBE2B	7320	GGACUGUUAGGAUUCGGUUca	AACCGAAUCCUAACAGUCC	-0.9	-3.3	2.7	2	0	3	-37.5	59.8	-1.6	3	III	47.4	39.9	49.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0224	NM_003337	Huesken	UBE2B	7320	GACUGUUAGGAUUCGGUUCau	GAACCGAAUCCUAACAGUC	-2.4	-2.4	1.9	2	0	3	-36.6	76	12.7	4	II	47.4	51.3	72.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0225	NM_003337	Huesken	UBE2B	7320	UUAGGAUUCGGUUCAUCCAgc	UGGAUGAACCGAAUCCUAA	-2.1	-0.9	-3.5	2	2	3	-36.6	75.1	-0.7	5	II	42.1	66.4	74.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0226	NM_003337	Huesken	UBE2B	7320	AGGAUUCGGUUCAUCCAGCag	GCUGGAUGAACCGAAUCCU	-3.4	-2.1	-3.5	2	3	3	-39.9	58.5	4.8	3	Ib	52.6	58.2	81.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0227	NM_003337	Huesken	UBE2B	7320	GGUUCAUCCAGCAGAGACUga	AGUCUCUGCUGGAUGAACC	-2.1	-3.3	-0.8	1	-1	2	-40.7	55.2	-8.9	4	II	52.6	38.1	53.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0228	NM_003337	Huesken	UBE2B	7320	UUCAUCCAGCAGAGACUGAau	UCAGUCUCUGCUGGAUGAA	-2.4	-0.9	-2.9	0	0	2	-39.7	58.8	-4	6	II	47.4	50.9	62.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0229	NM_003337	Huesken	UBE2B	7320	AUCCAGCAGAGACUGAAUUga	AAUUCAGUCUCUGCUGGAU	-0.9	-1.1	-4.8	3	1	2	-37.2	61	3.5	4	II	42.1	51.2	60.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0230	NM_003337	Huesken	UBE2B	7320	CUGAAUUGAUGUUAAGAUAga	UAUCUUAACAUCAAUUCAG	-1.3	-2.1	4	0	-4	1	-29.4	79.2	0	4	II	26.3	45.7	50.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0231	NM_003337	Huesken	UBE2B	7320	UUAAGAUAGAAGAUACAUCau	GAUGUAUCUUCUAUCUUAA	-2.4	-0.9	-0.1	2	4	1	-30.1	87	9.4	8	Ia	26.3	81.1	84.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0232	NM_003337	Huesken	UBE2B	7320	AUAGAAGAUACAUCAUAUGuu	CAUAUGAUGUAUCUUCUAU	-2.1	-1.1	1.2	1	3	1	-30.2	73.4	0.4	7	Ia	26.3	67.4	83.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0233	NM_003337	Huesken	UBE2B	7320	AUACAUCAUAUGUUGGACUcc	AGUCCAACAUAUGAUGUAU	-2.1	-1.1	0.3	2	3	2	-33.1	79.2	1.1	6	II	31.6	65.8	83.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0234	NM_003337	Huesken	UBE2B	7320	AUGUUGGACUCCAUCGAUUcu	AAUCGAUGGAGUCCAACAU	-0.9	-1.1	-0.3	2	1	2	-36.5	66.1	-8.9	6	II	42.1	56.6	74.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0235	NM_003337	Huesken	UBE2B	7320	UGUUGGACUCCAUCGAUUCug	GAAUCGAUGGAGUCCAACA	-2.4	-2.1	-0.3	0	2	2	-37.8	72	5.1	7	Ib	47.4	67.1	84.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0236	NM_003337	Huesken	UBE2B	7320	GUUGGACUCCAUCGAUUCUga	AGAAUCGAUGGAGUCCAAC	-2.1	-2.2	-0.3	0	0	2	-37.8	57.5	3.8	3	III	47.4	37	64.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0237	NM_003337	Huesken	UBE2B	7320	UUGGACUCCAUCGAUUCUGaa	CAGAAUCGAUGGAGUCCAA	-2.1	-0.9	-0.3	2	3	2	-37.7	67.8	0.4	7	Ib	47.4	74.5	86.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0238	NM_003337	Huesken	UBE2B	7320	GACUCCAUCGAUUCUGAAGga	CUUCAGAAUCGAUGGAGUC	-2.1	-2.4	-2.8	1	0	2	-36.8	78.3	5.3	2	II	47.4	44.7	72.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0239	NM_003337	Huesken	UBE2B	7320	CUCCAUCGAUUCUGAAGGAua	UCCUUCAGAAUCGAUGGAG	-2.4	-2.1	-2.5	-1	-2	2	-37.9	45.1	-5.4	1	III	47.4	31.7	50	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0240	NM_003337	Huesken	UBE2B	7320	UCCAUCGAUUCUGAAGGAUau	AUCCUUCAGAAUCGAUGGA	-1.1	-2.4	-0.8	0	0	2	-36.9	60.1	-3.3	6	II	42.1	46.7	77.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0241	NM_003337	Huesken	UBE2B	7320	CGAUUCUGAAGGAUAUCUAaa	UAGAUAUCCUUCAGAAUCG	-1.3	-2.4	0	0	-2	2	-33.8	71	-6	5	II	36.8	41.6	47.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0242	NM_003337	Huesken	UBE2B	7320	CUGAAGGAUAUCUAAACAUau	AUGUUUAGAUAUCCUUCAG	-1.1	-2.1	0	-1	-2	2	-31.8	64.2	-5.6	5	III	31.6	51.3	46.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0243	NM_003337	Huesken	UBE2B	7320	UGAAGGAUAUCUAAACAUAug	UAUGUUUAGAUAUCCUUCA	-1.3	-2.1	0	0	-1	2	-31	71.2	-5.7	6	II	26.3	58.3	63.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0244	NM_003337	Huesken	UBE2B	7320	UCUAAACAUAUGCUACCAUca	AUGGUAGCAUAUGUUUAGA	-1.1	-2.4	0.8	0	2	2	-33	71.8	3.5	7	II	31.6	54.8	69.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0245	NM_003337	Huesken	UBE2B	7320	UAUGCUACCAUCAGCAUACac	GUAUGCUGAUGGUAGCAUA	-2.2	-1.3	-5.4	0	3	2	-36.7	77.8	2.5	7	Ia	42.1	75.7	75.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0246	NM_003337	Huesken	UBE2B	7320	CCAUCAGCAUACACAUUUGga	CAAAUGUGUAUGCUGAUGG	-2.1	-3.3	1.2	-1	-1	2	-34.6	61.7	-4.6	4	II	42.1	43.3	62.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0247	NM_003337	Huesken	UBE2B	7320	UCAGCAUACACAUUUGGAUga	AUCCAAAUGUGUAUGCUGA	-1.1	-2.4	1.2	2	2	2	-34.9	74.6	-4	6	II	36.8	57.6	73	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0248	NM_003337	Huesken	UBE2B	7320	GCAUACACAUUUGGAUGAAac	UUCAUCCAAAUGUGUAUGC	-0.9	-3.4	1.2	-1	-2	2	-33.7	50.5	2	4	II	36.8	23.5	46	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0249	NM_003337	Huesken	UBE2B	7320	CACAUUUGGAUGAAACAUUuu	AAUGUUUCAUCCAAAUGUG	-0.9	-2.1	0.7	1	-3	2	-30.8	71.3	-8.6	5	II	31.6	53.7	63.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0250	NM_003337	Huesken	UBE2B	7320	UUUGGAUGAAACAUUUUGGau	CCAAAAUGUUUCAUCCAAA	-3.3	-0.9	2.2	1	5	2	-30.5	90.2	-0.3	8	Ia	31.6	78	84.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0251	NM_003337	Huesken	UBE2B	7320	GGAUGAAACAUUUUGGAUAaa	UAUCCAAAAUGUUUCAUCC	-1.3	-3.3	2.6	1	-2	2	-31.4	72.3	1.6	4	II	31.6	40.7	48.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0252	NM_003337	Huesken	UBE2B	7320	AAAACCUAACAGUUGGUGGuu	CCACCAACUGUUAGGUUUU	-3.3	-0.9	-1.3	2	5	2	-35	85.9	4.6	6	Ia	42.1	67.4	69.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0253	NM_003337	Huesken	UBE2B	7320	AGUUGGUGGUUUAUUUGGAua	UCCAAAUAAACCACCAACU	-2.4	-2.1		3	2	2	-34.2	69.7	-1	5	II	36.8	47.4	61.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0254	NM_003337	Huesken	UBE2B	7320	GUUUAUUUGGAUAUUCUUCag	GAAGAAUAUCCAAAUAAAC	-2.4	-2.2	1.9	2	2	2	-28.2	98.8	15.2	6	II	26.3	62.6	77.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0255	NM_003337	Huesken	UBE2B	7320	UUAUUUGGAUAUUCUUCAGaa	CUGAAGAAUAUCCAAAUAA	-2.1	-0.9	1.9	0	4	2	-29.3	100.7	5.3	8	Ia	26.3	71.9	80.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0256	NM_003337	Huesken	UBE2B	7320	UUUGGAUAUUCUUCAGAAAau	UUUCUGAAGAAUAUCCAAA	-0.9	-0.9	0.3	1	0	2	-30.2	81.1	-3.1	6	II	26.3	56.3	82.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0257	NM_003337	Huesken	UBE2B	7320	UUGGAUAUUCUUCAGAAAAuu	UUUUCUGAAGAAUAUCCAA	-0.9	-0.9	0.3	0	-1	2	-30.2	81.7	4.4	6	II	26.3	64.1	85.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0258	NM_003337	Huesken	UBE2B	7320	UGGAUAUUCUUCAGAAAAUuc	AUUUUCUGAAGAAUAUCCA	-1.1	-2.1	0.3	1	-1	2	-30.4	75.7	-8.9	6	II	26.3	56.2	63.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0259	NM_003337	Huesken	UBE2B	7320	UCAGAAAAUUCUAUUACUAgu	UAGUAAUAGAAUUUUCUGA	-1.3	-2.4	-0.8	0	1	1	-28.4	76.1	-0.6	6	II	21.1	58.8	72.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0260	NM_003337	Huesken	UBE2B	7320	AAAAUUCUAUUACUAGUUUaa	AAACUAGUAAUAGAAUUUU	-0.9	-0.9	-2	1	2	1	-25.5	93.2	0.8	6	II	15.8	61.1	47.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0261	NM_003337	Huesken	UBE2B	7320	AAUUCUAUUACUAGUUUAAaa	UUAAACUAGUAAUAGAAUU	-0.9	-0.9	-2	0	0	1	-25.9	99.5	-5.7	7	II	15.8	55.5	46.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0262	NM_003337	Huesken	UBE2B	7320	AUUACUAGUUUAAAAGUACca	GUACUUUUAAACUAGUAAU	-2.2	-1.1	-1.3	1	3	1	-26.8	86.3	4.8	6	Ia	21.1	69.8	86.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0263	NM_003337	Huesken	UBE2B	7320	AUCUUCAAAAGGUGUCCCUuc	AGGGACACCUUUUGAAGAU	-2.1	-1.1	-1.8	0	2	3	-36.7	74.8	-4	6	II	42.1	57.8	78.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0264	NM_003337	Huesken	UBE2B	7320	CUUCAAAAGGUGUCCCUUCug	GAAGGGACACCUUUUGAAG	-2.4	-2.1	-4	1	0	3	-36.5	58.8	8.1	4	II	47.4	55.6	79.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0265	NM_003337	Huesken	UBE2B	7320	UUCUGGUCCAAAUAUAACUgc	AGUUAUAUUUGGACCAGAA	-2.1	-0.9	0.8	2	1	2	-32.6	89.6	-1.7	7	II	31.6	75.2	79	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0266	NM_003337	Huesken	UBE2B	7320	UGGUCCAAAUAUAACUGCAuu	UGCAGUUAUAUUUGGACCA	-2.1	-2.1	0.1	0	0	2	-34.8	76	-1	5	II	36.8	56.8	81.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0267	NM_003337	Huesken	UBE2B	7320	GGUCCAAAUAUAACUGCAUuc	AUGCAGUUAUAUUUGGACC	-1.1	-3.3	0.1	1	-1	2	-33.8	56.9	-6.6	3	III	36.8	37.1	53.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0268	NM_003337	Huesken	UBE2B	7320	AAAUAUAACUGCAUUCCACug	GUGGAAUGCAGUUAUAUUU	-2.2	-0.9	0.7	2	5	2	-31.4	83.4	7.5	7	Ia	31.6	67.9	84.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0269	NM_003337	Huesken	UBE2B	7320	AUAUAACUGCAUUCCACUGca	CAGUGGAAUGCAGUUAUAU	-2.1	-1.1	0.5	2	4	2	-33.8	79.5	8	6	Ia	36.8	63.5	83.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0270	NM_003337	Huesken	UBE2B	7320	CAUUCCACUGCAUGAUGUUgu	AACAUCAUGCAGUGGAAUG	-0.9	-2.1	-0.1	-2	-1	2	-35.7	63.8	-8.3	4	II	42.1	40.7	55	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0271	NM_003337	Huesken	UBE2B	7320	UCCACUGCAUGAUGUUGUUuu	AACAACAUCAUGCAGUGGA	-0.9	-2.4	-0.1	-1	0	2	-36.8	67.9	-4	5	II	42.1	47.1	56.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0272	NM_003337	Huesken	UBE2B	7320	GCAUGAUGUUGUUUUCAGAug	UCUGAAAACAACAUCAUGC	-2.4	-3.4	0.6	1	-1	2	-33.4	69.2	1.6	3	II	36.8	34.4	48.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0273	NM_003337	Huesken	UBE2B	7320	GAUGUUGUUUUCAGAUGGUgc	ACCAUCUGAAAACAACAUC	-2.2	-2.4	2	-1	1	2	-33.4	60.5	-8.9	4	II	36.8	43	50.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0274	NM_003337	Huesken	UBE2B	7320	UGUUGUUUUCAGAUGGUGCgc	GCACCAUCUGAAAACAACA	-3.4	-2.1	2	1	4	2	-35.4	88.5	9.4	7	Ia	42.1	72.5	74.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0275	NM_003337	Huesken	UBE2B	7320	UUGUUUUCAGAUGGUGCGCca	GCGCACCAUCUGAAAACAA	-3.4	-0.9	2	0	4	4	-36.9	79	12.7	7	Ia	47.4	72.2	72.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0276	NM_003337	Huesken	UBE2B	7320	UUCAGAUGGUGCGCCACUGac	CAGUGGCGCACCAUCUGAA	-2.1	-0.9	-2.2	2	3	5	-41.7	55.1	-2.4	4	Ia	57.9	63.7	61.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0277	NM_003337	Huesken	UBE2B	7320	GCCACUGACACCCACAGGUgg	ACCUGUGGGUGUCAGUGGC	-2.2	-3.4	-1.9	1	-1	3	-44.7	43.7	-6.3	1	III	63.2	37.3	76.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0278	NM_003337	Huesken	UBE2B	7320	CACAGGUGGGUCCUCUUGUaa	ACAAGAGGACCCACCUGUG	-2.2	-2.1	-1.5	1	-1	3	-42.4	52.3	-3.6	2	III	57.9	45.7	61.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0279	NM_003337	Huesken	UBE2B	7320	CUCUUGUAACCGCUUGAAAuc	UUUCAAGCGGUUACAAGAG	-0.9	-2.1	2.3	0	-3	4	-34.6	67.4	-3.4	4	II	42.1	36.2	45.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0280	NM_003337	Huesken	UBE2B	7320	UCUUGUAACCGCUUGAAAUcc	AUUUCAAGCGGUUACAAGA	-1.1	-2.4	2.3	1	1	4	-33.6	83.3	-4	6	II	36.8	58.1	78.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0281	NM_003337	Huesken	UBE2B	7320	GUAACCGCUUGAAAUCCCGca	CGGGAUUUCAAGCGGUUAC	-2.4	-2.2	0.8	0	3	4	-37.5	55.1	-2.6	4	II	52.6	54.8	85.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0282	NM_003969	Huesken	UBE2M	9040	GGAGCCGAUGUAGCCACCCcg	GGGUGGCUACAUCGGCUCC	-3.3	-3.3	-3.5	0	2	4	-45.7	31.7	7.4	0	II	68.4	47.8	58.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0283	NM_003969	Huesken	UBE2M	9040	AGCCGAUGUAGCCACCCCGca	CGGGGUGGCUACAUCGGCU	-2.4	-2.1	-3.5	3	3	5	-45.7	41.5	0.1	2	II	68.4	55.9	54.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0284	NM_003969	Huesken	UBE2M	9040	GAUGUAGCCACCCCGCAUGga	CAUGCGGGGUGGCUACAUC	-2.1	-2.4	-1.8	1	1	6	-43.2	48.8	-2.3	3	II	63.2	47.5	67.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0285	NM_003969	Huesken	UBE2M	9040	GCAUGGAGCGCUGCACGUUcu	AACGUGCAGCGCUCCAUGC	-0.9	-3.4	-1	1	-1	4	-43.1	31.2	-3.3	1	III	63.2	31.5	61.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0286	NM_003969	Huesken	UBE2M	9040	UUCUGCUCAAACAGCCGCCgg	GGCGGCUGUUUGAGCAGAA	-3.3	-0.9	-3	0	4	6	-41.5	65.9	4.8	5	Ib	57.9	72.7	76.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0287	NM_003969	Huesken	UBE2M	9040	AACAGCCGCCGGUUGUUCUgc	AGAACAACCGGCGGCUGUU	-2.1	-0.9	0.8	2	2	8	-41.6	63.6	-1.6	4	II	57.9	49.4	61.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0288	NM_003969	Huesken	UBE2M	9040	AGCCGCCGGUUGUUCUGCAgg	UGCAGAACAACCGGCGGCU	-2.1	-2.1	0	3	1	8	-44	40.3	-1.4	1	II	63.2	35.1	67.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0289	NM_003969	Huesken	UBE2M	9040	GCCGCCGGUUGUUCUGCAGga	CUGCAGAACAACCGGCGGC	-2.1	-3.4	-2.2	0	0	8	-44	42.6	3	-1	II	68.4	27.7	64.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0290	NM_003969	Huesken	UBE2M	9040	UGUUCUGCAGGACCUCUGCgg	GCAGAGGUCCUGCAGAACA	-3.4	-2.1	-5.2	0	3	2	-42.7	69.9	9.8	5	Ib	57.9	64	69.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0291	NM_003969	Huesken	UBE2M	9040	UUCUGCAGGACCUCUGCGGcc	CCGCAGAGGUCCUGCAGAA	-3.3	-0.9	-5.5	1	4	4	-44.1	61.8	-4.7	5	Ib	63.2	67.6	67.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0292	NM_003969	Huesken	UBE2M	9040	CUGCAGGACCUCUGCGGCCuc	GGCCGCAGAGGUCCUGCAG	-3.3	-2.1	-5.5	0	1	6	-47.5	29.2	5.8	1	II	73.7	49	66.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0293	NM_003969	Huesken	UBE2M	9040	UGCAGGACCUCUGCGGCCUcc	AGGCCGCAGAGGUCCUGCA	-2.1	-2.1	-5.5	1	1	6	-47.5	44.3	-3.3	3	II	68.4	53.3	65.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0294	NM_003969	Huesken	UBE2M	9040	UCUGCGGCCUCCUUGUUCAgu	UGAACAAGGAGGCCGCAGA	-2.1	-2.4	0.5	1	1	6	-42.9	61.2	-6.1	5	II	57.9	53.1	65.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0295	NM_003969	Huesken	UBE2M	9040	CUGCGGCCUCCUUGUUCAGug	CUGAACAAGGAGGCCGCAG	-2.1	-2.1	0.5	-1	0	6	-42.6	40.7	-1.3	0	II	63.2	35.6	68.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0296	NM_003969	Huesken	UBE2M	9040	CUCCUUGUUCAGUGGGUCCuc	GGACCCACUGAACAAGGAG	-3.3	-2.1	0.2	-1	0	3	-41.4	61	7.7	3	II	57.9	52.3	63.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0297	NM_003969	Huesken	UBE2M	9040	CAGUGGGUCCUCGGGGUUGgg	CAACCCCGAGGACCCACUG	-2.1	-2.1	-0.3	-1	-1	5	-45	34.2	-4.4	3	II	68.4	42.2	62.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0298	NM_003969	Huesken	UBE2M	9040	CUCGGGGUUGGGCUCCAAGaa	CUUGGAGCCCAACCCCGAG	-2.1	-2.1	-3.2	-1	0	5	-44.9	39.9	2.7	0	II	68.4	42.8	64.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0299	NM_003969	Huesken	UBE2M	9040	GGGGUUGGGCUCCAAGAAGag	CUUCUUGGAGCCCAACCCC	-2.1	-3.3	-0.7	1	-1	4	-43.4	33.8	0.1	1	II	63.2	35.5	61.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0300	NM_003969	Huesken	UBE2M	9040	GGGUUGGGCUCCAAGAAGAga	UCUUCUUGGAGCCCAACCC	-2.4	-3.3	-0.7	1	-3	4	-42.5	38.2	-11	2	III	57.9	31.3	56.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0301	NM_003969	Huesken	UBE2M	9040	GUUGGGCUCCAAGAAGAGAua	UCUCUUCUUGGAGCCCAAC	-2.4	-2.2	-2.3	1	-1	4	-40.4	38.6	-1.5	1	III	52.6	33.9	69.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0302	NM_003969	Huesken	UBE2M	9040	UUGGGCUCCAAGAAGAGAUac	AUCUCUUCUUGGAGCCCAA	-1.1	-0.9	-2.3	2	1	4	-39.3	65.5	-4.3	5	II	47.4	63.8	77.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0303	NM_003969	Huesken	UBE2M	9040	AGAAGAGAUACUGCAGGCCau	GGCCUGCAGUAUCUCUUCU	-3.3	-2.1	1.5	1	4	4	-40.8	50.9	8.1	6	Ib	52.6	59.8	80.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0304	NM_003969	Huesken	UBE2M	9040	UACUGCAGGCCAUAAAUUAug	UAAUUUAUGGCCUGCAGUA	-1.3	-1.3	1.5	1	-1	4	-34.9	69.4	-6.1	7	II	36.8	56.7	67	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0305	NM_003969	Huesken	UBE2M	9040	GGCCAUAAAUUAUGGAGUUua	AACUCCAUAAUUUAUGGCC	-0.9	-3.3	-2.1	0	-2	4	-33.7	52.9	-1.3	1	III	36.8	32.3	49.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0306	NM_003969	Huesken	UBE2M	9040	AAAUUAUGGAGUUUAUCGUaa	ACGAUAAACUCCAUAAUUU	-2.2	-0.9	1.8	3	4	2	-29.5	94.4	1.4	8	II	26.3	64.4	82.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0307	NM_003969	Huesken	UBE2M	9040	UGGAGUUUAUCGUAAGGACug	GUCCUUACGAUAAACUCCA	-2.2	-2.1	0.7	1	2	2	-35.5	73.4	5.1	5	Ib	42.1	61.6	76.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0308	NM_003969	Huesken	UBE2M	9040	GGAGUUUAUCGUAAGGACUgg	AGUCCUUACGAUAAACUCC	-2.1	-3.3	0.4	1	-1	2	-35.5	64.1	-3.5	3	II	42.1	47.5	60.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0309	NM_003969	Huesken	UBE2M	9040	GUUUAUCGUAAGGACUGGCuu	GCCAGUCCUUACGAUAAAC	-3.4	-2.2	0.4	0	3	3	-36.5	64.2	12.4	5	II	47.4	52.4	68.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0310	NM_003969	Huesken	UBE2M	9040	AGGACUGGCUUCCAGUCCUcu	AGGACUGGAAGCCAGUCCU	-2.1	-2.1	-11.3	2	1	3	-43.8	51.8	-3.9	2	II	57.9	51.2	54.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0311	NM_003969	Huesken	UBE2M	9040	AGGAUGUUGAGGCAGACGUug	ACGUCUGCCUCAACAUCCU	-2.2	-2.1	1.2	3	1	3	-40.8	60.9	-1.6	6	II	52.6	55.3	80.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0312	NM_003969	Huesken	UBE2M	9040	UUGAGGCAGACGUUGCCCUcg	AGGGCAACGUCUGCCUCAA	-2.1	-0.9	-5.7	1	3	4	-42.7	58.5	-4	4	II	57.9	64.3	79.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0313	NM_003969	Huesken	UBE2M	9040	CGAGGUCAAUGUUGGGGUGau	CACCCCAACAUUGACCUCG	-2.1	-2.4	1.8	-2	0	4	-40.4	43.8	1	0	II	57.9	34.4	62.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0314	NM_003969	Huesken	UBE2M	9040	GAGGUCAAUGUUGGGGUGAua	UCACCCCAACAUUGACCUC	-2.4	-2.4	1.7	-1	-2	4	-40.4	44.6	-0.7	4	III	52.6	35.7	65.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0315	NM_003969	Huesken	UBE2M	9040	AAUGUUGGGGUGAUAGACCau	GGUCUAUCACCCCAACAUU	-3.3	-0.9	-0.8	2	5	4	-38.3	64.3	7.1	6	Ia	47.4	70.1	79.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0316	NM_003969	Huesken	UBE2M	9040	UGUUGGGGUGAUAGACCAUug	AUGGUCUAUCACCCCAACA	-1.1	-2.1	-0.8	-1	1	4	-39.5	53.8	-1.2	6	II	47.4	48.8	71.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0317	NM_003969	Huesken	UBE2M	9040	UUGGGGUGAUAGACCAUUGuc	CAAUGGUCUAUCACCCCAA	-2.1	-0.9	-0.8	1	2	4	-38.2	65.2	-0.3	6	II	47.4	70.2	86.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0318	NM_003969	Huesken	UBE2M	9040	GGGGUGAUAGACCAUUGUCuc	GACAAUGGUCUAUCACCCC	-2.4	-3.3	-1.6	0	0	4	-39.8	53.1	9.8	1	II	52.6	44	61.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0319	NM_003969	Huesken	UBE2M	9040	UAGACCAUUGUCUCACACUuc	AGUGUGAGACAAUGGUCUA	-2.1	-1.3	-3.1	0	1	2	-37.3	77.7	-6.3	5	II	42.1	70.5	92.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0320	NM_003969	Huesken	UBE2M	9040	CAUUGUCUCACACUUCACCuu	GGUGAAGUGUGAGACAAUG	-3.3	-2.1	0.8	0	2	2	-36.9	75.2	7.8	4	II	47.4	59.4	81.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0321	NM_003969	Huesken	UBE2M	9040	CAUGCGGGUAACCCUGGCCca	GGCCAGGGUUACCCGCAUG	-3.3	-2.1	-2.8	-1	2	5	-45.1	48.1	7.8	2	II	68.4	53.5	73.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0322	NM_003969	Huesken	UBE2M	9040	AUGCGGGUAACCCUGGCCCac	GGGCCAGGGUUACCCGCAU	-3.3	-1.1	-3	2	5	5	-46.3	48.5	7.4	3	II	68.4	64.4	79.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0323	NM_003969	Huesken	UBE2M	9040	GCGGGUAACCCUGGCCCACcu	GUGGGCCAGGGUUACCCGC	-2.2	-3.4	-5.2	0	0	5	-47.4	23.4	8.1	-1	II	73.7	32.8	56.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0324	NM_003969	Huesken	UBE2M	9040	GUAACCCUGGCCCACCUUAaa	UAAGGUGGGCCAGGGUUAC	-1.3	-2.2	-1.8	1	-2	5	-42.6	48.3	-6.1	2	III	57.9	36.3	69.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0325	NM_003969	Huesken	UBE2M	9040	GCCCACCUUAAAACUGAACac	GUUCAGUUUUAAGGUGGGC	-2.2	-3.4	-0.4	1	-1	4	-36.5	54.4	9.8	2	II	47.4	36.5	53.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0326	NM_003969	Huesken	UBE2M	9040	CCUUAAAACUGAACACAAAcu	UUUGUGUUCAGUUUUAAGG	-0.9	-3.3	1.1	0	-4	2	-30.4	57.4	-5.6	4	II	31.6	31.7	51.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0327	NM_003969	Huesken	UBE2M	9040	ACACAAACUUCCCACUCUUgu	AAGAGUGGGAAGUUUGUGU	-0.9	-2.2	2	2	1	3	-36.4	52.1	-3.6	5	II	42.1	48.6	85.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0328	NM_003969	Huesken	UBE2M	9040	AAACUUCCCACUCUUGUAGaa	CUACAAGAGUGGGAAGUUU	-2.1	-0.9	-2.5	3	4	3	-35.5	82.6	5.4	6	Ia	42.1	62.1	87.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0329	NM_003969	Huesken	UBE2M	9040	ACUUCCCACUCUUGUAGAAgc	UUCUACAAGAGUGGGAAGU	-0.9	-2.2	-4.3	1	0	3	-37	65.4	-3.1	3	II	42.1	31.2	68.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0330	NM_003969	Huesken	UBE2M	9040	CUUCCCACUCUUGUAGAAGcc	CUUCUACAAGAGUGGGAAG	-2.1	-2.1	-4.4	0	1	3	-36.9	63.8	3.4	2	II	47.4	48.2	75.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0331	NM_003969	Huesken	UBE2M	9040	UCUUGUAGAAGCCCUCAUCag	GAUGAGGGCUUCUACAAGA	-2.4	-2.4	0.9	0	2	4	-38.6	74.7	7.5	7	Ia	47.4	66.8	82.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0332	NM_003969	Huesken	UBE2M	9040	UGUAGAAGCCCUCAUCAGGac	CCUGAUGAGGGCUUCUACA	-3.3	-2.1	-1.8	1	3	4	-40.7	62	3.1	5	Ia	52.6	63.2	77.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0333	NM_003969	Huesken	UBE2M	9040	UAGAAGCCCUCAUCAGGACag	GUCCUGAUGAGGGCUUCUA	-2.2	-1.3	-5.7	1	3	4	-41	58.9	7.8	5	Ib	52.6	59.8	80.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0334	NM_003969	Huesken	UBE2M	9040	CCCUCAUCAGGACAGAUGAcc	UCAUCUGUCCUGAUGAGGG	-2.4	-3.3	-4.9	0	-4	3	-41	52.8	-3.4	2	III	52.6	32.5	57.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0335	NM_003969	Huesken	UBE2M	9040	CCUCAUCAGGACAGAUGACca	GUCAUCUGUCCUGAUGAGG	-2.2	-3.3	-5.1	0	0	2	-39.9	40.9	5.5	1	II	52.6	34.3	58.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0336	NM_003969	Huesken	UBE2M	9040	CAUCAGGACAGAUGACCAGcu	CUGGUCAUCUGUCCUGAUG	-2.1	-2.1	-2.3	0	1	2	-39.6	46.4	0.7	4	II	52.6	45.3	76.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0337	NM_003969	Huesken	UBE2M	9040	ACAGAUGACCAGCUUGAAGuu	CUUCAAGCUGGUCAUCUGU	-2.1	-2.2	1.6	2	3	2	-38	60.2	9.7	5	Ib	47.4	53	77	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0338	NM_003969	Huesken	UBE2M	9040	GUCGUCUGGAUCUGAGAAGcu	CUUCUCAGAUCCAGACGAC	-2.1	-2.2	1.1	1	0	2	-39.1	53.8	-2.4	4	II	52.6	43.2	65.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0339	NM_003969	Huesken	UBE2M	9040	GUCUGGAUCUGAGAAGCUGau	CAGCUUCUCAGAUCCAGAC	-2.1	-2.2	1.1	1	1	2	-39.7	47.5	0	3	II	52.6	44.7	79.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0340	NM_003969	Huesken	UBE2M	9040	GAUCUGAGAAGCUGAUAUCac	GAUAUCAGCUUCUCAGAUC	-2.4	-2.4	0.9	0	1	2	-35.9	65.3	5.1	5	II	42.1	49.4	60.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0341	NM_003969	Huesken	UBE2M	9040	UCUGAGAAGCUGAUAUCACac	GUGAUAUCAGCUUCUCAGA	-2.2	-2.4	0.9	1	4	2	-36.7	60.4	9.8	6	Ib	42.1	63.4	70.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0342	NM_003969	Huesken	UBE2M	9040	GUCUUGGGCAGGUUCAGCUcg	AGCUGAACCUGCCCAAGAC	-2.1	-2.2	-0.5	1	1	4	-42.4	61.1	-1.6	4	III	57.9	47.2	71.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0343	NM_003969	Huesken	UBE2M	9040	CUUGGGCAGGUUCAGCUCGuu	CGAGCUGAACCUGCCCAAG	-2.4	-2.1	-0.5	0	1	4	-42.6	43.2	3.4	2	II	63.2	51.8	71.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0344	NM_003969	Huesken	UBE2M	9040	UUGGGCAGGUUCAGCUCGUuu	ACGAGCUGAACCUGCCCAA	-2.2	-0.9	-0.5	0	2	4	-42.7	42.3	-8.9	4	II	57.9	57.6	66	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0345	NM_003969	Huesken	UBE2M	9040	CGUUUAUGUCCUUCUGGAUcc	AUCCAGAAGGACAUAAACG	-1.1	-2.4	-1.9	0	-2	2	-35.2	71	-5.3	3	II	42.1	34.8	56.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0346	NM_003969	Huesken	UBE2M	9040	GAUCCGCAGCUGCGCCGCCga	GGCGGCGCAGCUGCGGAUC	-3.3	-2.4	-6.3	1	3	8	-48.4	24.7	9.8	0	II	78.9	38.7	71.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0347	NM_003969	Huesken	UBE2M	9040	AUCCGCAGCUGCGCCGCCGac	CGGCGGCGCAGCUGCGGAU	-2.4	-1.1	-6.3	3	4	9	-48.4	33.5	-2.4	2	II	78.9	54.9	69.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0348	NM_003969	Huesken	UBE2M	9040	AGCUGCGCCGCCGACGCCUuc	AGGCGUCGGCGGCGCAGCU	-2.1	-2.1	-6.1	2	1	9	-49.5	27.6	-8.7	2	II	78.9	43.7	51.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0349	NM_003969	Huesken	UBE2M	9040	CCGACGCCUUCUUGCUGCUgc	AGCAGCAAGAAGGCGUCGG	-2.1	-3.3	-1.3	-2	-2	4	-43	41.9	-5.3	-1	III	63.2	30.2	53	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0350	NM_003969	Huesken	UBE2M	9040	GACGCCUUCUUGCUGCUGCcc	GCAGCAGCAAGAAGGCGUC	-3.4	-2.4	-1.3	2	2	4	-42.8	60.3	12.1	1	II	63.2	48.4	61	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0351	NM_003969	Huesken	UBE2M	9040	CGCCUUCUUGCUGCUGCCCuu	GGGCAGCAGCAAGAAGGCG	-3.3	-2.4	-1.3	0	0	4	-44.8	45.7	6.1	-1	II	68.4	42.9	58.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0352	NM_003969	Huesken	UBE2M	9040	GCUGCCCUUGGUGCCGCCCgc	GGGCGGCACCAAGGGCAGC	-3.3	-3.4	-2	0	2	7	-49.4	36.4	10.4	-1	II	78.9	38.2	64.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0353	NM_003969	Huesken	UBE2M	9040	UUGGUGCCGCCCGCCGACUcc	AGUCGGCGGGCGGCACCAA	-2.1	-0.9	-4.1	1	1	11	-47.9	36.2	-8.7	3	II	73.7	56.2	73.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0354	NM_003969	Huesken	UBE2M	9040	UGGUGCCGCCCGCCGACUCcu	GAGUCGGCGGGCGGCACCA	-2.4	-2.1	-4.1	1	2	11	-49.4	39.2	7.5	2	II	78.9	51.5	53	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0355	NM_003969	Huesken	UBE2M	9040	UGCCGCCCGCCGACUCCUCcu	GAGGAGUCGGCGGGCGGCA	-2.4	-2.1	-4.1	2	2	11	-49.6	31.2	2.5	1	II	78.9	48.6	59.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0356	NM_003969	Huesken	UBE2M	9040	CCCGCCGACUCCUCCUCCUuc	AGGAGGAGGAGUCGGCGGG	-2.1	-3.3	1.3	0	-2	7	-48.2	35	-10.7	0	III	73.7	33	41.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0357	NM_003969	Huesken	UBE2M	9040	GACUCCUCCUCCUUCUUCUgc	AGAAGAAGGAGGAGGAGUC	-2.1	-2.4	3	2	0	2	-40.9	75.3	-6.3	4	III	52.6	44.7	59.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0358	XM_371822	Huesken	C6orf110	55362	ACUGGAAGUCUGUGUCCGUga	ACGGACACAGACUUCCAGU	-2.2	-2.2	-2.3	0	2	3	-40.7	42	-4	3	II	52.6	43.8	85.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0359	XM_371822	Huesken	C6orf110	55362	GUCUGUGUCCGUGAGGGCGuc	CGCCCUCACGGACACAGAC	-2.4	-2.2	-0.5	1	2	5	-44.6	47.7	3	2	II	68.4	46.3	59.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0360	XM_371822	Huesken	C6orf110	55362	CUCAUCUUGGGAUGAGGAUga	AUCCUCAUCCCAAGAUGAG	-1.1	-2.1	-5.8	0	-2	3	-38.7	59.6	-6	3	II	47.4	37.1	53.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0361	XM_371822	Huesken	C6orf110	55362	GGGGCUCAUCCCCUGAGCUcc	AGCUCAGGGGAUGAGCCCC	-2.1	-3.3	-6.5	0	0	5	-47.5	38.9	-3.9	-2	III	68.4	32	43	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0362	XM_371822	Huesken	C6orf110	55362	CCUGAGCUCCCAGGAGCCCca	GGGCUCCUGGGAGCUCAGG	-3.3	-3.3	-10.5	-1	1	4	-48.5	20.1	5.8	0	II	73.7	37.3	36.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0363	XM_371822	Huesken	C6orf110	55362	CUCCCAGGAGCCCCAUCCCca	GGGAUGGGGCUCCUGGGAG	-3.3	-2.1	-2.9	0	1	5	-48.6	34.6	3.1	0	II	73.7	44.9	43.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0364	XM_371822	Huesken	C6orf110	55362	CAUCCCCGUCCACCUCUGAgu	UCAGAGGUGGACGGGGAUG	-2.4	-2.1	1.5	0	-2	5	-44.5	43	-5.7	2	III	63.2	31.4	29.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0365	XM_371822	Huesken	C6orf110	55362	UCUGAGUCCUGCAGCACCUga	AGGUGCUGCAGGACUCAGA	-2.1	-2.4	-0.1	1	2	2	-43.9	49.4	-6.2	5	II	57.9	58.4	61.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0366	XM_371822	Huesken	C6orf110	55362	AUUUCGCAGAUUUGGGGACag	GUCCCCAAAUCUGCGAAAU	-2.2	-1.1	2.3	1	4	4	-37.2	68.5	10.4	6	Ib	47.4	55	82	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0367	XM_371822	Huesken	C6orf110	55362	UUGCUUCUGGGGUCCACAGua	CUGUGGACCCCAGAAGCAA	-2.1	-0.9	-0.4	2	3	4	-42.3	65.2	0	4	Ib	57.9	63.6	75.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0368	XM_371822	Huesken	C6orf110	55362	UCUUGUAGUUGUGGGCACUga	AGUGCCCACAACUACAAGA	-2.1	-2.4	0	-1	1	4	-39	62.7	-1	6	II	47.4	54.7	81.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0369	XM_371822	Huesken	C6orf110	55362	CACUGAGGUAUUUGAAGUGuc	CACUUCAAAUACCUCAGUG	-2.1	-2.1	0.1	-2	0	2	-34.5	59.8	1	4	II	42.1	47	55	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0370	XM_371822	Huesken	C6orf110	55362	CAGACGUGGCAGAGACAGAug	UCUGUCUCUGCCACGUCUG	-2.4	-2.1	1.9	0	-3	3	-42.1	38.8	-6	2	III	57.9	35.7	68.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0371	XM_371822	Huesken	C6orf110	55362	ACAUAGACGUGGGAGCUAGga	CUAGCUCCCACGUCUAUGU	-2.1	-2.2	1.6	2	2	3	-40.1	48.8	2.7	7	Ia	52.6	49.8	66.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0372	XM_371822	Huesken	C6orf110	55362	AUAGACGUGGGAGCUAGGAac	UCCUAGCUCCCACGUCUAU	-2.4	-1.1	1.6	2	2	3	-41.5	45.5	-1.1	6	II	52.6	49.9	79.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0373	XM_371822	Huesken	C6orf110	55362	ACGUGGGAGCUAGGAACCCcg	GGGUUCCUAGCUCCCACGU	-3.3	-2.2	-3.4	1	3	3	-44.3	27.3	7.4	2	II	63.2	51.2	57.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0374	XM_371822	Huesken	C6orf110	55362	CAGAAGAGGCAGAGGAUGGgc	CCAUCCUCUGCCUCUUCUG	-3.3	-2.1	2.5	-1	0	3	-41.7	43.4	0.4	4	II	57.9	49.3	53.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0375	XM_371822	Huesken	C6orf110	55362	CUUCUUGUCCAGCUUGGCCgg	GGCCAAGCUGGACAAGAAG	-3.3	-2.1	-0.8	1	3	4	-41.1	67.4	12.4	3	II	57.9	57.4	65	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0376	XM_371822	Huesken	C6orf110	55362	UGUCCAGCUUGGCCGGCAGgu	CUGCCGGCCAAGCUGGACA	-2.1	-2.1	-5.8	1	3	7	-46	46.8	2.4	3	II	68.4	48.6	65.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0377	XM_371822	Huesken	C6orf110	55362	CUACCAGGUGCUUCAGCAGca	CUGCUGAAGCACCUGGUAG	-2.1	-2.1	-2	0	1	2	-41.3	48.1	-1.3	1	II	57.9	44.5	75	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0378	XM_371822	Huesken	C6orf110	55362	UACCAGGUGCUUCAGCAGCau	GCUGCUGAAGCACCUGGUA	-3.4	-1.3	-3	2	3	2	-42.6	55.5	15.5	4	II	57.9	67	82.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0379	XM_371822	Huesken	C6orf110	55362	AUGACCACCGUGAAGACGCac	GCGUCUUCACGGUGGUCAU	-3.4	-1.1	-0.6	2	4	3	-41.2	57.5	4.8	4	Ib	57.9	69.3	75.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0380	XM_371822	Huesken	C6orf110	55362	GCGUAGGCUGCGCCAAACUgg	AGUUUGGCGCAGCCUACGC	-2.1	-3.4	-2.2	0	-2	5	-43	35	-1.2	1	III	63.2	34.5	60.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0381	XM_371822	Huesken	C6orf110	55362	UGAUGCCGCUUCACGUUGCgc	GCAACGUGAAGCGGCAUCA	-3.4	-2.1	-1	1	3	5	-40.9	58.4	2.5	4	II	57.9	61.2	72.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0382	XM_371822	Huesken	C6orf110	55362	GCUUCACGUUGCGCCUCUCgg	GAGAGGCGCAACGUGAAGC	-2.4	-3.4	0.2	0	1	5	-42.2	43.3	5.1	1	II	63.2	32	56.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0383	XM_371822	Huesken	C6orf110	55362	UUGCGCCUCUCGGCGGCCGag	CGGCCGCCGAGAGGCGCAA	-2.4	-0.9	-8.4	2	4	9	-48.4	37.2	-2.4	2	II	78.9	58.6	72.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0384	XM_371822	Huesken	C6orf110	55362	GGAUCAUGUACAUGAGCAGgc	CUGCUCAUGUACAUGAUCC	-2.1	-3.3	-1.4	0	1	2	-37.6	54.5	-2.4	3	II	47.4	38.7	49	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0385	XM_371822	Huesken	C6orf110	55362	UCAUGUACAUGAGCAGGCCug	GGCCUGCUCAUGUACAUGA	-3.3	-2.4	-0.8	0	4	4	-40.8	57.9	7.4	6	Ia	52.6	65.3	71.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0386	XM_371822	Huesken	C6orf110	55362	AGCAGGUCCAUGGCGUUGCcg	GCAACGCCAUGGACCUGCU	-3.4	-2.1	-0.5	3	2	4	-43.9	52.8	7.4	5	II	63.2	58.4	71.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0387	XM_371822	Huesken	C6orf110	55362	UCCAUGGCGUUGCCGAUAAag	UUAUCGGCAACGCCAUGGA	-0.9	-2.4	-3	1	-2	4	-40.1	54.1	-1.4	6	II	52.6	44.9	62.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0388	XM_371822	Huesken	C6orf110	55362	GAAAGUGUAGCACUUGUGCau	GCACAAGUGCUACACUUUC	-3.4	-2.4	-2.6	0	3	2	-36.7	66.4	9.8	4	II	47.4	55.4	83.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0389	XM_371822	Huesken	C6orf110	55362	AAAGUGUAGCACUUGUGCAug	UGCACAAGUGCUACACUUU	-2.1	-0.9	-2.6	3	3	2	-36.4	71.4	-1.5	5	II	42.1	59.2	99.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0390	XM_371822	Huesken	C6orf110	55362	GGAGGGCCGAGAAGCACCAca	UGGUGCUUCUCGGCCCUCC	-2.1	-3.3	-1.3	0	-1	6	-46.5	18.7	-6.3	0	III	68.4	30.4	65.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0391	XM_371822	Huesken	C6orf110	55362	GGUGGGGAAGAACUGGGUGau	CACCCAGUUCUUCCCCACC	-2.1	-3.3	2.3	0	2	4	-43.5	37	7	1	II	63.2	37.4	56.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0392	XM_371822	Huesken	C6orf110	55362	UCCAUGGUGGUGAUGAUGAug	UCAUCAUCACCACCAUGGA	-2.4	-2.4	-1.1	1	0	2	-39.8	62	-4	6	II	47.4	56.3	79.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0393	XM_371822	Huesken	C6orf110	55362	CCAUGGUGGUGAUGAUGAUgg	AUCAUCAUCACCACCAUGG	-1.1	-3.3	0	0	-2	2	-38.5	44.5	-6	3	III	47.4	27.7	59.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0394	XM_371822	Huesken	C6orf110	55362	AUGAUGAUGGCUGGAGUGGug	CCACUCCAGCCAUCAUCAU	-3.3	-1.1	4.6	1	3	3	-40.7	56.1	0.7	7	Ia	52.6	60.2	77.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0395	XM_371822	Huesken	C6orf110	55362	GAUGAUGGCUGGAGUGGUGag	CACCACUCCAGCCAUCAUC	-2.1	-2.4		0	2	3	-41.8	42.5	0.1	3	II	57.9	40	70.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0396	XM_371822	Huesken	C6orf110	55362	UGGUGAGGAAGAAGAGGAGga	CUCCUCUUCUUCCUCACCA	-2.1	-2.1		-1	2	2	-40.6	44.9	-2.6	6	II	52.6	50.1	68	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0397	XM_371822	Huesken	C6orf110	55362	AGGAGGAUGAAGAGGACGAca	UCGUCCUCUUCAUCCUCCU	-2.4	-2.1	4.5	1	0	2	-41.4	44.9	-4	5	II	52.6	46.7	55.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0398	XM_371822	Huesken	C6orf110	55362	AGAUGAAGCCUCGGAUGGAga	UCCAUCCGAGGCUUCAUCU	-2.4	-2.1	0.1	1	1	3	-41.3	35.8	-6.1	5	II	52.6	38.5	65.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0399	XM_371822	Huesken	C6orf110	55362	GGCGUGGCUCCCCACGGCAgg	UGCCGUGGGGAGCCACGCC	-2.1	-3.3	-8	0	-2	4	-50	19.5	-6.1	0	III	78.9	27.3	62.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0400	XM_371822	Huesken	C6orf110	55362	AUUUACACACGUUGAAGUCcu	GACUUCAACGUGUGUAAAU	-2.4	-1.1	0.8	0	4	2	-32.6	80.3	13.1	7	Ia	36.8	61.5	101.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0401	XM_371822	Huesken	C6orf110	55362	CACGUUGAAGUCCUUCAGGau	CCUGAAGGACUUCAACGUG	-3.3	-2.1	-1.4	-1	1	2	-38.1	58.7	0.4	2	II	52.6	53	84.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0402	XM_371822	Huesken	C6orf110	55362	AUAGUCUCAUUGUGGAAGGug	CCUUCCACAAUGAGACUAU	-3.3	-1.1	1.1	1	4	2	-36.1	70.7	0	6	Ia	42.1	65	72.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0403	XM_371822	Huesken	C6orf110	55362	UUGUGGAAGGUGACAAAGGcc	CCUUUGUCACCUUCCACAA	-3.3	-0.9	0.6	1	3	2	-37.5	59	-2.6	6	Ib	47.4	72.8	74.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0404	XM_371822	Huesken	C6orf110	55362	UCAAUGGCCUCCACCUGCUca	AGCAGGUGGAGGCCAUUGA	-2.1	-2.4	-2.4	1	2	4	-43.7	52.8	-11.3	5	II	57.9	60.6	77.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0405	XM_371822	Huesken	C6orf110	55362	CCACCUGCUCACAGCCUCGca	CGAGGCUGUGAGCAGGUGG	-2.4	-3.3	-2.3	-1	0	3	-45.1	37.6	-4.6	1	II	68.4	46.2	42.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0406	XM_371822	Huesken	C6orf110	55362	UGAUCAUGGUAGGCACGUUcu	AACGUGCCUACCAUGAUCA	-0.9	-2.1	-1.6	2	1	3	-38.7	58.5	-1.7	5	II	47.4	50.5	80.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0407	XM_371822	Huesken	C6orf110	55362	GGAGGUUUGUGAAGUACAGcu	CUGUACUUCACAAACCUCC	-2.1	-3.3	0.6	1	1	2	-36.8	48.8	-2.6	2	II	47.4	39.5	57.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0408	XM_371822	Huesken	C6orf110	55362	CUCGGCCUUCUUCCUCUCUgc	AGAGAGGAAGAAGGCCGAG	-2.1	-2.1	-0.6	-1	-2	5	-42.1	53.5	-0.6	1	III	57.9	41.6	49.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0409	XM_371822	Huesken	C6orf110	55362	CUUCUUCCUCUCUGCAUCGag	CGAUGCAGAGAGGAAGAAG	-2.4	-2.1		-1	1	2	-38.7	62.7	-4.4	4	II	52.6	55.2	56.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0410	XM_371822	Huesken	C6orf110	55362	AUCGAGGAACAUUAGGCGAgc	UCGCCUAAUGUUCCUCGAU	-2.4	-1.1	1.3	1	1	4	-38.2	56.1	4.3	5	II	47.4	49.8	77.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0411	XM_371822	Huesken	C6orf110	55362	CGAGCCACGUUGUAACACGgg	CGUGUUACAACGUGGCUCG	-2.4	-2.4	-1.7	0	0	3	-38.8	46.1	-2	0	II	57.9	49	91.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0412	XM_371822	Huesken	C6orf110	55362	AGCCACGUUGUAACACGGGcg	CCCGUGUUACAACGUGGCU	-3.3	-2.1	-2.4	2	3	4	-40.6	40.8	0.1	2	II	57.9	52.2	75.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0413	XM_371822	Huesken	C6orf110	55362	UGUGCAGUUGGGGUAGGCUuc	AGCCUACCCCAACUGCACA	-2.1	-2.1	0.5	0	3	4	-43.5	52.2	-1.6	4	II	57.9	53	70.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0414	XM_371822	Huesken	C6orf110	55362	CAAAAUGCUUCUUGAUCUUuu	AAGAUCAAGAAGCAUUUUG	-0.9	-2.1	0.3	-2	-1	2	-30.8	70.1	-2.6	5	II	31.6	45.9	61.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0415	XM_371822	Huesken	C6orf110	55362	GCAUAUUUGGAGAUUCCAUug	AUGGAAUCUCCAAAUAUGC	-1.1	-3.4	-2.5	2	0	2	-34	69.5	0.8	3	II	36.8	41	78.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0416	XM_371822	Huesken	C6orf110	55362	GAUCAUCCUCCUUGUAGCGca	CGCUACAAGGAGGAUGAUC	-2.4	-2.4	1.2	0	3	3	-39.1	58.9	3.4	3	II	52.6	45.6	79.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0417	XM_371822	Huesken	C6orf110	55362	UGGAGGUGUGUCUACGCAUgc	AUGCGUAGACACACCUCCA	-1.1	-2.1	1.1	1	0	3	-41	54.2	-6.3	5	II	52.6	56.6	76.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0418	XM_371822	Huesken	C6orf110	55362	GACGGUGAGCAGCAGAUACag	GUAUCUGCUGCUCACCGUC	-2.2	-2.4	0.6	1	0	3	-41.3	47.5	12.1	2	II	57.9	42.3	65.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0419	XM_371822	Huesken	C6orf110	55362	GUUGUUCCCUGAUUUCAAGuu	CUUGAAAUCAGGGAACAAC	-2.1	-2.2	-2.6	1	2	3	-34.3	74.5	2.3	2	II	42.1	46.2	71.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0420	XM_371822	Huesken	C6orf110	55362	AUGGUGGUUCUCCCAAAGCug	GCUUUGGGAGAACCACCAU	-3.4	-1.1	-2.5	1	3	3	-40.1	57	7.5	4	II	52.6	65.5	75.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0421	XM_371822	Huesken	C6orf110	55362	GUUCUCCCAAAGCUGUAGGca	CCUACAGCUUUGGGAGAAC	-3.3	-2.2	-2.3	1	3	3	-39.1	68.4	7	3	II	52.6	48.1	81.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0422	XM_371822	Huesken	C6orf110	55362	GCACGAUGCCUACGGAGAGga	CUCUCCGUAGGCAUCGUGC	-2.1	-3.4	0.4	2	1	3	-42.7	27.4	2.4	1	II	63.2	31.9	75.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0423	XM_371822	Huesken	C6orf110	55362	CAGCAGCCCGAUGAUGUGCcg	GCACAUCAUCGGGCUGCUG	-3.4	-2.1	-0.3	1	0	5	-43.1	44.6	13.4	2	II	63.2	47.2	69.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0424	XM_371822	Huesken	C6orf110	55362	AAGGACAGGUAGUGCACGGca	CCGUGCACUACCUGUCCUU	-3.3	-0.9	1	1	4	3	-41.7	48.5	5	4	Ib	57.9	56.6	86.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0425	XM_371822	Huesken	C6orf110	55362	GUCCCGGAUCUCAUCAUCCuu	GGAUGAUGAGAUCCGGGAC	-3.3	-2.2	-3.8	1	2	5	-41.8	52.3	4.8	1	II	57.9	54.8	89.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0426	XM_371822	Huesken	C6orf110	55362	CGGAUCUCAUCAUCCUUUAuc	UAAAGGAUGAUGAGAUCCG	-1.3	-2.4	-4	0	-5	3	-35.8	67.5	-8.1	3	III	42.1	36.8	24.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0427	XM_371822	Huesken	C6orf110	55362	UGUCCCUUUGGUCAAAGUCaa	GACUUUGACCAAAGGGACA	-2.4	-2.1	-1.3	2	3	3	-37.8	76.3	12.8	4	II	47.4	65.1	89.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0428	XM_371822	Huesken	C6orf110	55362	UUGGUCAAAGUCAACGGAGcu	CUCCGUUGACUUUGACCAA	-2.1	-0.9	0.1	0	3	3	-36.9	60.8	-4.9	6	Ib	47.4	69.6	98.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0429	XM_371822	Huesken	C6orf110	55362	AGUCAACGGAGCUGGAGACag	GUCUCCAGCUCCGUUGACU	-2.2	-2.1	-1.1	2	3	3	-42.1	45.7	7.8	3	Ib	57.9	42.7	69.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0430	XM_371822	Huesken	C6orf110	55362	CUGGAGACAGAGGUGAGACgc	GUCUCACCUCUGUCUCCAG	-2.2	-2.1	0.3	-1	1	2	-42	46.4	12.7	2	II	57.9	48.5	88.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0431	XM_371822	Huesken	C6orf110	55362	UGAGACGCUCAUACCGGUCau	GACCGGUAUGAGCGUCUCA	-2.4	-2.1	-0.7	0	3	4	-41.8	55.6	12.8	5	II	57.9	63.8	99	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0432	XM_371822	Huesken	C6orf110	55362	UCAUACCGGUCAUGGCUGUcc	ACAGCCAUGACCGGUAUGA	-2.2	-2.4	-2.4	0	1	4	-41.1	51.5	-3.6	6	II	52.6	46.8	56.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0433	XM_371822	Huesken	C6orf110	55362	GGCUGUCCCCGUGCAUAGCug	GCUAUGCACGGGGACAGCC	-3.4	-3.3	0.1	1	0	5	-45.5	44.2	10.4	0	II	68.4	33.5	73.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0434	XM_371822	Huesken	C6orf110	55362	GGAAGUCUGUGUCCGUGAGgg	CUCACGGACACAGACUUCC	-2.1	-3.3	-0.6	1	1	3	-40.9	52.6	5.4	1	II	57.9	32.7	61.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0435	XM_371822	Huesken	C6orf110	55362	GUGAGGGCGUCGGGUGGCAuc	UGCCACCCGACGCCCUCAC	-2.1	-2.2	0.5	0	-1	5	-47.9	16.8	-6.1	2	III	73.7	26.6	12.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0436	XM_371822	Huesken	C6orf110	55362	CGUCGGGUGGCAUCAGCAAcu	UUGCUGAUGCCACCCGACG	-0.9	-2.4	-0.6	1	-3	4	-43.2	29.1	-8.1	1	III	63.2	25.4	39.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0437	XM_371822	Huesken	C6orf110	55362	AACUCCUCAUCUUGGGAUGag	CAUCCCAAGAUGAGGAGUU	-2.1	-0.9	-2.1	1	2	3	-38.3	75.2	0.7	4	Ib	47.4	53	53	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0438	XM_371822	Huesken	C6orf110	55362	AUCCCCUGAGCUCCCAGGAgc	UCCUGGGAGCUCAGGGGAU	-2.4	-1.1	-6.3	3	1	4	-46.5	52.1	-0.7	2	II	63.2	46.1	52	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0439	XM_371822	Huesken	C6orf110	55362	GUCCUGCAGCACCUGAGCGau	CGCUCAGGUGCUGCAGGAC	-2.4	-2.2	-1.5	2	3	3	-45.2	43.7	4.7	0	II	68.4	43.7	46.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0440	XM_371822	Huesken	C6orf110	55362	CCUGAGCGAUGUAUUUCGCag	GCGAAAUACAUCGCUCAGG	-3.4	-3.3	-4.3	-1	2	3	-38	53.4	10.9	1	II	52.6	41.8	40.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0441	XM_371822	Huesken	C6orf110	55362	AGCGAUGUAUUUCGCAGAUuu	AUCUGCGAAAUACAUCGCU	-1.1	-2.1	-4.3	0	0	3	-35.8	56.8	4.2	3	II	42.1	40.2	50.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0442	XM_371822	Huesken	C6orf110	55362	UUCUGGGGUCCACAGUAUCug	GAUACUGUGGACCCCAGAA	-2.4	-0.9	-0.9	0	2	4	-40.8	58.1	7.5	6	II	52.6	64.8	73.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0443	XM_371822	Huesken	C6orf110	55362	AGUUGUGGGCACUGAGGUAuu	UACCUCAGUGCCCACAACU	-1.3	-2.1	0.3	1	0	4	-41.4	47.8	-3.8	5	II	52.6	36.4	51.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0444	XM_371822	Huesken	C6orf110	55362	UGACGAUGGUGAUGACCAGga	CUGGUCAUCACCAUCGUCA	-2.1	-2.1	-2.3	2	3	2	-40	46.4	0	4	Ib	52.6	51.4	67	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0445	XM_371822	Huesken	C6orf110	55362	CGUGCGCAUGGUGGAAAAGaa	CUUUUCCACCAUGCGCACG	-2.1	-2.4	2.8	-2	-1	4	-39.3	38.3	1	0	II	57.9	31.1	0	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0446	XM_371822	Huesken	C6orf110	55362	GCGCAUGGUGGAAAAGAAGag	CUUCUUUUCCACCAUGCGC	-2.1	-3.4	3.1	1	-1	4	-38	36.1	0.1	2	II	52.6	36.6	39.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0447	XM_371822	Huesken	C6orf110	55362	GAGGAUGGGCGCGGCCACCac	GGUGGCCGCGCCCAUCCUC	-3.3	-2.4	-4.4	0	1	10	-49.2	21.4	7.8	1	II	78.9	42.3	49.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0448	XM_371822	Huesken	C6orf110	55362	GGAUGGGCGCGGCCACCACcu	GUGGUGGCCGCGCCCAUCC	-2.2	-3.3	-4.4	1	1	10	-49	23.9	9.8	1	II	78.9	31.9	46.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0449	XM_371822	Huesken	C6orf110	55362	GGCGCGGCCACCACCUGGUuc	ACCAGGUGGUGGCCGCGCC	-2.2	-3.3	-3.2	1	-1	9	-49.8	25.3	-11.3	0	III	78.9	31.1	57.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0450	XM_371822	Huesken	C6orf110	55362	GCAGGUAGGCGUAGUAGAGau	CUCUACUACGCCUACCUGC	-2.1	-3.4	0.9	0	1	4	-41.3	37.2	0.4	1	II	57.9	34.7	56.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0451	XM_371822	Huesken	C6orf110	55362	ACGAUGAUGGGGCAGGUGAua	UCACCUGCCCCAUCAUCGU	-2.4	-2.2	2.6	2	-1	5	-43.3	48.4	-1.4	5	II	57.9	45.1	73	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0452	XM_371822	Huesken	C6orf110	55362	GGUGAUACUGUAGGUCAUGac	CAUGACCUACAGUAUCACC	-2.1	-3.3	1.5	-1	0	2	-37.6	45.1	2.7	3	II	47.4	42.7	58.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0453	XM_371822	Huesken	C6orf110	55362	GUGAUACUGUAGGUCAUGAcc	UCAUGACCUACAGUAUCAC	-2.4	-2.2	1.5	1	-1	2	-36.7	62.4	0.9	4	II	42.1	37.1	60.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0454	XM_371822	Huesken	C6orf110	55362	ACAUCAUCCAGGCGUAGGCug	GCCUACGCCUGGAUGAUGU	-3.4	-2.2	-0.7	2	4	4	-42.3	58.8	9.8	5	Ia	57.9	55.1	72.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0455	XM_371822	Huesken	C6orf110	55362	CCGCUUCACGUUGCGCCUCuc	GAGGCGCAACGUGAAGCGG	-2.4	-3.3	-3.2	1	-1	5	-43.4	37.3	8.4	-1	II	68.4	41.4	59.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0456	XM_371822	Huesken	C6orf110	55362	CGCUUCACGUUGCGCCUCUcg	AGAGGCGCAACGUGAAGCG	-2.1	-2.4	-1.9	-1	-3	5	-42.2	43.9	-3.6	2	III	63.2	34.9	35.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0457	XM_371822	Huesken	C6orf110	55362	ACGUUGCGCCUCUCGGCGGcc	CCGCCGAGAGGCGCAACGU	-3.3	-2.2	-8.4	2	3	6	-46.1	40.5	-2.4	2	II	73.7	49.1	21.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0458	XM_371822	Huesken	C6orf110	55362	GCAGAGCCGGAUCAUGUACau	GUACAUGAUCCGGCUCUGC	-2.2	-3.4	1.5	1	0	5	-41.4	43.6	17.8	2	II	57.9	36.4	46.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0459	XM_371822	Huesken	C6orf110	55362	GAUCAUGUACAUGAGCAGGcc	CCUGCUCAUGUACAUGAUC	-3.3	-2.4	-1.4	1	2	2	-37.6	61.2	8	3	II	47.4	50.1	73.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0460	XM_371822	Huesken	C6orf110	55362	AUGUACAUGAGCAGGCCUGgg	CAGGCCUGCUCAUGUACAU	-2.1	-1.1	0	1	3	4	-40.5	58.8	-2.2	7	Ia	52.6	61.2	72.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0461	XM_371822	Huesken	C6orf110	55362	UGAGCAGGCCUGGGAUGCGca	CGCAUCCCAGGCCUGCUCA	-2.4	-2.1	-0.9	0	4	4	-46.4	31.5	0	4	II	68.4	54.6	74.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0462	XM_371822	Huesken	C6orf110	55362	GCAGCAGGUCCAUGGCGUUgc	AACGCCAUGGACCUGCUGC	-0.9	-3.4	0.7	-1	-1	4	-43.9	25.3	-6.3	1	III	63.2	24.8	27.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0463	XM_371822	Huesken	C6orf110	55362	CCAUGGCGUUGCCGAUAAAgg	UUUAUCGGCAACGCCAUGG	-0.9	-3.3	-3	-2	-4	4	-38.6	34.5	-5.7	1	III	52.6	18	36.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0464	XM_371822	Huesken	C6orf110	55362	ACCAUGAAGAUGAGGAAAGug	CUUUCCUCAUCUUCAUGGU	-2.1	-2.2	0.5	1	1	2	-35.9	56.4	-2.6	6	Ib	42.1	54.1	72.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0465	XM_371822	Huesken	C6orf110	55362	CUUGUGCAUGGUUGUCCUGuu	CAGGACAACCAUGCACAAG	-2.1	-2.1	-0.7	-2	1	2	-38.7	56.3	1	3	II	52.6	46.4	61.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0466	XM_371822	Huesken	C6orf110	55362	AUGGUUGUCCUGUUCUCCCca	GGGAGAACAGGACAACCAU	-3.3	-1.1	-0.7	2	5	3	-40.4	72.5	9.7	5	Ib	52.6	72.6	65.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0467	XM_371822	Huesken	C6orf110	55362	UUGUCCUGUUCUCCCCAGAgc	UCUGGGGAGAACAGGACAA	-2.4	-0.9	-2.5	1	0	4	-41.6	70.4	-0.7	5	II	52.6	57.4	68	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0468	XM_371822	Huesken	C6orf110	55362	CCUGUUCUCCCCAGAGCGUgu	ACGCUCUGGGGAGAACAGG	-2.2	-3.3	-2.1	-1	-1	4	-44.1	36	-13.3	0	III	63.2	32.6	48.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0469	XM_371822	Huesken	C6orf110	55362	AGCACCACAGCAGCAGGGUgg	ACCCUGCUGCUGUGGUGCU	-2.2	-2.1	-1.6	1	1	3	-45.6	40	-6.3	2	II	63.2	41.5	58.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0470	XM_371822	Huesken	C6orf110	55362	AGAACUGGGUGAUGAUGGGgu	CCCAUCAUCACCCAGUUCU	-3.3	-2.1	3.6	1	4	3	-40.5	52.4	0	5	Ib	52.6	50	68.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0471	XM_371822	Huesken	C6orf110	55362	AACUUGUCCAUGGUGGUGAug	UCACCACCAUGGACAAGUU	-2.4	-0.9	-1.1	3	1	2	-39	70.5	-1.4	7	II	47.4	56.1	67.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0472	XM_371822	Huesken	C6orf110	55362	GUGGUGAUGAUGAUGGCUGga	CAGCCAUCAUCAUCACCAC	-2.1	-2.2	4.5	1	2	3	-39.6	55.1	-0.3	4	II	52.6	55.5	74.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0473	XM_371822	Huesken	C6orf110	55362	AAGAAGAGGAGGAUGAAGAgg	UCUUCAUCCUCCUCUUCUU	-2.4	-0.9		2	1	2	-37.4	62.5	-1.3	7	II	42.1	57.3	58.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0474	XM_371822	Huesken	C6orf110	55362	UGAAGAGGACGACAUUGAUga	AUCAAUGUCGUCCUCUUCA	-1.1	-2.1	3.1	0	1	2	-36.6	56.4	-1.6	6	II	42.1	48.4	49.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0475	XM_371822	Huesken	C6orf110	55362	GACGACAUUGAUGACCAGGca	CCUGGUCAUCAAUGUCGUC	-3.3	-2.4	1.9	0	1	2	-38.8	54.3	8	3	II	52.6	50.1	60.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0476	XM_371822	Huesken	C6orf110	55362	UGAUGACCAGGCAGCGCAGcc	CUGCGCUGCCUGGUCAUCA	-2.1	-2.1	0.4	-1	3	4	-44.1	40.9	-4.9	4	Ib	63.2	49.4	66.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0477	XM_371822	Huesken	C6orf110	55362	GCAGCGCAGCCACCAGAUGaa	CAUCUGGUGGCUGCGCUGC	-2.1	-3.4	-2.2	1	0	4	-45.1	24.6	0.1	-1	II	68.4	31.4	29.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0478	XM_371822	Huesken	C6orf110	55362	CACCAGAUGAAGCCUCGGAug	UCCGAGGCUUCAUCUGGUG	-2.4	-2.1	-0.6	1	-2	3	-42.1	45.6	1.7	3	III	57.9	38.5	29.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0479	XM_371822	Huesken	C6orf110	55362	UGUGCAGGGACUCGCUGCAgg	UGCAGCGAGUCCCUGCACA	-2.1	-2.1	-4.5	0	1	3	-45.2	44.5	-0.7	4	II	63.2	46.7	25.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0480	XM_371822	Huesken	C6orf110	55362	CAGCCCUGGCAUUUACACAcg	UGUGUAAAUGCCAGGGCUG	-2.1	-2.1	-1.9	2	-2	4	-40	62.2	1.9	1	III	52.6	42	66.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0481	XM_371822	Huesken	C6orf110	55362	GCCCUGGCAUUUACACACGuu	CGUGUGUAAAUGCCAGGGC	-2.4	-3.4	-0.7	1	0	4	-40.4	42.2	0.4	1	II	57.9	41.3	52.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0482	XM_371822	Huesken	C6orf110	55362	UGUGGAAGGUGACAAAGGCca	GCCUUUGUCACCUUCCACA	-3.4	-2.1	1.5	1	4	3	-40	49.2	9.8	5	Ib	52.6	60.1	72.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0483	XM_371822	Huesken	C6orf110	55362	CCAAGAGGCUUCUCAUUCAcc	UGAAUGAGAAGCCUCUUGG	-2.1	-3.3	-1.6	0	-3	3	-38.1	48.8	-8.1	3	III	47.4	37.6	39.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0484	XM_371822	Huesken	C6orf110	55362	CAAGAGGCUUCUCAUUCACcu	GUGAAUGAGAAGCCUCUUG	-2.2	-2.1	-1.6	-1	1	3	-37	55.4	11.2	3	II	47.4	49.2	69.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0485	XM_371822	Huesken	C6orf110	55362	UCCUUCUCCCGCUUGUAGUcu	ACUACAAGCGGGAGAAGGA	-2.2	-2.4	1.2	2	1	5	-40.9	78.5	-4	6	II	52.6	56.2	72	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0486	XM_371822	Huesken	C6orf110	55362	CUCCAGCUUUGUGUAGUACuc	GUACUACACAAAGCUGGAG	-2.2	-2.1	0.7	0	0	2	-37	60.2	13.5	3	II	47.4	41.2	58.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0487	XM_371822	Huesken	C6orf110	55362	GGGGUUGAUCAUGGUAGGCac	GCCUACCAUGAUCAACCCC	-3.4	-3.3	4.7	-1	1	4	-41.9	44.8	12.7	1	II	57.9	39.8	39.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0488	XM_371822	Huesken	C6orf110	55362	UGGUAGGCACGUUCUCCUUgc	AAGGAGAACGUGCCUACCA	-0.9	-2.1	-1.6	0	0	3	-40.7	63	1.8	5	II	52.6	52.2	70.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0489	XM_371822	Huesken	C6orf110	55362	GGGUAGGCUUCCUCAAAAUgc	AUUUUGAGGAAGCCUACCC	-1.1	-3.3	-0.3	0	-3	3	-38	50.2	-6	3	III	47.4	29.7	27.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0490	XM_371822	Huesken	C6orf110	55362	GUAGGCUUCCUCAAAAUGCuu	GCAUUUUGAGGAAGCCUAC	-3.4	-2.2	-0.3	2	2	3	-36.9	59.6	2.5	4	II	47.4	59.9	87.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0491	XM_371822	Huesken	C6orf110	55362	GAUUCCAUUGAUGAAGAGGgu	CCUCUUCAUCAAUGGAAUC	-3.3	-2.4	-0.1	0	2	2	-35.1	72.9	5.3	4	II	42.1	52.1	73.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0492	XM_371822	Huesken	C6orf110	55362	CCGCUUCACCAGAUCAUCCuc	GGAUGAUCUGGUGAAGCGG	-3.3	-3.3	0.4	1	0	4	-41.1	55.9	7.4	1	II	57.9	50.6	63.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0493	XM_371822	Huesken	C6orf110	55362	UUCACCAGAUCAUCCUCCUug	AGGAGGAUGAUCUGGUGAA	-2.1	-0.9	-0.7	0	2	2	-39.8	72.7	-6.3	6	II	47.4	64.2	77.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0494	XM_371822	Huesken	C6orf110	55362	GAGCAGCAGAUACAGGAAGgc	CUUCCUGUAUCUGCUGCUC	-2.1	-2.4	0.6	2	0	2	-39.6	40.5	5.1	3	II	52.6	41.2	75.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0495	XM_371822	Huesken	C6orf110	55362	UUGUUCCCUGAUUUCAAGUug	ACUUGAAAUCAGGGAACAA	-2.2	-0.9	-3.1	0	2	3	-34.3	92.6	3.7	5	II	36.8	67.2	34.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0496	XM_371822	Huesken	C6orf110	55362	CAAGUUGGCAAUGGUGGUUcu	AACCACCAUUGCCAACUUG	-0.9	-2.1	2.6	-1	-1	3	-37.2	55.7	-0.7	5	II	47.4	44.3	34.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0497	XM_371822	Huesken	C6orf110	55362	GGCAAUGGUGGUUCUCCCAaa	UGGGAGAACCACCAUUGCC	-2.1	-3.3	-2.3	0	-2	3	-42.5	51	2	1	III	57.9	34.2	43.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0498	XM_371822	Huesken	C6orf110	55362	GAGAAGUUGACAGGCAGCAcg	UGCUGCCUGUCAACUUCUC	-2.1	-2.4	1.3	1	-1	3	-40.3	42.7	-3.4	5	II	52.6	41.6	60.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0499	XM_371822	Huesken	C6orf110	55362	AGAGGACGCCCACAACCACca	GUGGUUGUGGGCGUCCUCU	-2.2	-2.1	-0.1	2	3	5	-44	29.9	7.5	2	II	63.2	44.8	53.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0500	XM_371822	Huesken	C6orf110	55362	GAUGUGCCGCUGAAAGGACag	GUCCUUUCAGCGGCACAUC	-2.2	-2.4	-1	1	2	5	-40.8	39.7	4.8	1	II	57.9	41.1	65.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0501	XM_371822	Huesken	C6orf110	55362	ACCAUUGUCCCUUUGGUCAaa	UGACCAAAGGGACAAUGGU	-2.1	-2.2	-1.4	2	0	3	-38.9	75.9	-0.8	5	II	47.4	51	70.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0502	XM_371822	Huesken	C6orf110	55362	GUCAAAGUCAACGGAGCUGga	CAGCUCCGUUGACUUUGAC	-2.1	-2.2	-0.6	0	1	3	-38.2	47.4	2.7	5	II	52.6	44.1	84.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0503	NM_015213	Huesken	RAB6IP1	23258	CAGCACACGAAUCAAGGAAuu	UUCCUUGAUUCGUGUGCUG	-0.9	-2.1	2.8	1	-3	2	-37.1	52.6	4.7	4	III	47.4	36.4	22.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0504	NM_015213	Huesken	RAB6IP1	23258	AGCCAGCAGGGCAAUCCAGug	CUGGAUUGCCCUGCUGGCU	-2.1	-2.1	-3.1	3	2	4	-44.6	33	-2.2	1	II	63.2	42.6	63.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0505	NM_015213	Huesken	RAB6IP1	23258	UAAUGUCUCAUAAUAGGUUug	AACCUAUUAUGAGACAUUA	-0.9	-1.3	1.3	1	3	2	-30.8	84.3	-4.2	7	II	26.3	69.6	66.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0506	NM_015213	Huesken	RAB6IP1	23258	CUCCACAGAGCAACAGCGUca	ACGCUGUUGCUCUGUGGAG	-2.2	-2.1	-1.3	0	-1	3	-41.6	37.2	-8.2	1	III	57.9	41.1	60.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0507	NM_015213	Huesken	RAB6IP1	23258	CUCAGGCUUAUGGAAGUGCuu	GCACUUCCAUAAGCCUGAG	-3.4	-2.1	0.8	-1	0	3	-39.3	51.1	5.4	4	II	52.6	52.4	78.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0508	NM_015213	Huesken	RAB6IP1	23258	GCUGGGAUGUGAGCAGCUCcc	GAGCUGCUCACAUCCCAGC	-2.4	-3.4	-2.1	1	1	3	-44.1	31.2	7.4	0	II	63.2	37.1	42.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0509	NM_015213	Huesken	RAB6IP1	23258	CACAUACUCCACCAGCCAUuu	AUGGCUGGUGGAGUAUGUG	-1.1	-2.1	-0.3	0	-2	3	-40.5	55	-3.6	3	II	52.6	36.2	71.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0510	NM_015213	Huesken	RAB6IP1	23258	GUAAGCUUCCCCAAGUUCUgg	AGAACUUGGGGAAGCUUAC	-2.1	-2.2	-1.1	2	0	4	-37.9	64.3	-11.3	4	II	47.4	52.7	88.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0511	NM_015213	Huesken	RAB6IP1	23258	CUUCCCCAAGUUCUGGCACuc	GUGCCAGAACUUGGGGAAG	-2.2	-2.1	-1.3	0	2	4	-41.1	58.7	13.2	0	II	57.9	44.5	41.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0512	NM_015213	Huesken	RAB6IP1	23258	UGUCUCACCCAAUUCUCCUga	AGGAGAAUUGGGUGAGACA	-2.1	-2.1	1.7	2	3	3	-39.4	76.9	0.7	6	II	47.4	64.6	70.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0513	NM_015213	Huesken	RAB6IP1	23258	UCUUGCUUGGUACGAUCAGaa	CUGAUCGUACCAAGCAAGA	-2.1	-2.4	0.6	1	3	2	-37.5	61.7	2.4	6	Ib	47.4	56.3	54.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0514	NM_015213	Huesken	RAB6IP1	23258	UAAUCGACGGCAUUGAAAGac	CUUUCAAUGCCGUCGAUUA	-2.1	-1.3	2.2	1	4	4	-34.3	78.1	0	7	Ia	42.1	70.9	76.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0515	NM_015213	Huesken	RAB6IP1	23258	CGACGGCAUUGAAAGACAGga	CUGUCUUUCAAUGCCGUCG	-2.1	-2.4	1.6	0	0	4	-37.4	36	-4.6	1	II	52.6	38.3	47.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0516	NM_015213	Huesken	RAB6IP1	23258	UCCCCGAUGUUCUGGAUGUgc	ACAUCCAGAACAUCGGGGA	-2.2	-2.4	-1.3	1	0	5	-41.1	55.6	-6.3	4	II	52.6	50.9	44.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0517	NM_015213	Huesken	RAB6IP1	23258	GAAUCAGGGAGAUCCUCAGgg	CUGAGGAUCUCCCUGAUUC	-2.1	-2.4	-2.9	1	2	3	-40	57.7	0	4	II	52.6	45.1	60.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0518	NM_015213	Huesken	RAB6IP1	23258	GGGACCAUAAGGCUGAUUUcc	AAAUCAGCCUUAUGGUCCC	-0.9	-3.3	-2.1	0	-2	3	-38.2	69.5	0.8	3	III	47.4	39.2	41.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0519	NM_015213	Huesken	RAB6IP1	23258	ACCAUAAGGCUGAUUUCCCcu	GGGAAAUCAGCCUUAUGGU	-3.3	-2.2	-2	2	4	3	-38.2	62.1	7.1	4	Ia	47.4	60.5	61.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0520	NM_015213	Huesken	RAB6IP1	23258	UUAUUGCGGCAUUCCUUCAgc	UGAAGGAAUGCCGCAAUAA	-2.1	-0.9	-0.9	1	1	5	-36.1	80.7	1.6	8	II	42.1	63.2	77.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0521	NM_015213	Huesken	RAB6IP1	23258	ACAGUUUGGGCUGGCGGAUag	AUCCGCCAGCCCAAACUGU	-1.1	-2.2	-0.8	2	1	5	-42.6	41.4	-3.3	4	II	57.9	41.3	40.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0522	NM_015213	Huesken	RAB6IP1	23258	UGGAUGUACUUCUCCCUCUgg	AGAGGGAGAAGUACAUCCA	-2.1	-2.1	-2.9	2	0	3	-39.8	72.6	-6.3	6	II	47.4	66.3	80.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0523	NM_015213	Huesken	RAB6IP1	23258	CACUUGUUGCUGGCUGGCCca	GGCCAGCCAGCAACAAGUG	-3.3	-2.1	-1.8	1	1	4	-43.1	52.7	5.4	2	II	63.2	52	73.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0524	NM_015213	Huesken	RAB6IP1	23258	CCUUGUCCAAUCUUCAUGUca	ACAUGAAGAUUGGACAAGG	-2.2	-3.3	0.9	-1	-1	2	-35.7	73	-6	4	III	42.1	45.9	70.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0525	NM_015213	Huesken	RAB6IP1	23258	GAAUUGCAGUAUGGUCAAUuu	AUUGACCAUACUGCAAUUC	-1.1	-2.4	1.4	0	-1	2	-33.7	59.1	1.4	4	II	36.8	31.3	45.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0526	NM_015213	Huesken	RAB6IP1	23258	GAUGUACGGAGAGUAGGUGuc	CACCUACUCUCCGUACAUC	-2.1	-2.4	0.6	1	3	3	-39.2	48	2.3	4	II	52.6	41.6	54.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0527	NM_015213	Huesken	RAB6IP1	23258	GUAGGUGUCCGAACAUUCAac	UGAAUGUUCGGACACCUAC	-2.1	-2.2	0.5	1	-1	3	-37.6	53.1	-6.3	4	III	47.4	45	49	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0528	NM_015213	Huesken	RAB6IP1	23258	UUAUUUUGUUGUCAAUGAAag	UUCAUUGACAACAAAAUAA	-0.9	-0.9	0.9	1	1	1	-27.4	95.3	1.7	8	II	21.1	62	97.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0529	NM_015213	Huesken	RAB6IP1	23258	UGAAAGAUGCAAACAUCUGgg	CAGAUGUUUGCAUCUUUCA	-2.1	-2.1	-4.2	1	3	2	-33.3	68.1	2.4	6	Ia	36.8	69.2	75.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0530	NM_015213	Huesken	RAB6IP1	23258	AGAUGCAAACAUCUGGGUCuc	GACCCAGAUGUUUGCAUCU	-2.4	-2.1	-4.1	1	4	3	-38.2	70	17.5	4	Ib	47.4	56.6	71.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0531	NM_015213	Huesken	RAB6IP1	23258	CUUGAGAGGAAGGGCAGGUag	ACCUGCCCUUCCUCUCAAG	-2.2	-2.1	2	-1	0	4	-42.5	37.5	-1	4	II	57.9	43.7	53.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0532	NM_015213	Huesken	RAB6IP1	23258	GUUUUGCAUUUGCUCCCUGuu	CAGGGAGCAAAUGCAAAAC	-2.1	-2.2	-1.5	0	3	3	-36.2	63.9	4.7	3	II	47.4	48.7	67.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0533	NM_015213	Huesken	RAB6IP1	23258	UUUGCAUUUGCUCCCUGUUgg	AACAGGGAGCAAAUGCAAA	-0.9	-0.9	-1.5	1	2	3	-36.2	79.3	-0.9	5	II	42.1	63.5	49.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0534	NM_015213	Huesken	RAB6IP1	23258	UAUCCUGGCUGGGUUGGAUga	AUCCAACCCAGCCAGGAUA	-1.1	-1.3	-2	2	3	3	-41.8	63.3	-1.6	5	II	52.6	51.5	51.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0535	NM_015213	Huesken	RAB6IP1	23258	GUAAAUCCUGAGUUCUUCUuc	AGAAGAACUCAGGAUUUAC	-2.1	-2.2	0.8	0	1	2	-33.8	79.8	3.4	5	II	36.8	53.1	44.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0536	NM_015213	Huesken	RAB6IP1	23258	UUUCCAGGCUCACCCCAGUuc	ACUGGGGUGAGCCUGGAAA	-2.2	-0.9	-4.6	2	2	4	-43.5	52.6	-3.9	4	II	57.9	57	37.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0537	NM_015213	Huesken	RAB6IP1	23258	UUCCAGGCUCACCCCAGUUcu	AACUGGGGUGAGCCUGGAA	-0.9	-0.9	-4.6	1	1	4	-43.5	52.2	1.1	4	II	57.9	57.4	58.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0538	NM_015213	Huesken	RAB6IP1	23258	CAAGGCUUGCAGCCGGGCAau	UGCCCGGCUGCAAGCCUUG	-2.1	-2.1	-4.3	1	-1	7	-45.7	40	-1.1	1	III	68.4	39.8	56.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0539	NM_015213	Huesken	RAB6IP1	23258	CGUAGGAAUGCAAAGGGGAgc	UCCCCUUUGCAUUCCUACG	-2.4	-2.4	2.8	-1	-2	4	-39.5	37.1	-10.7	3	III	52.6	37.7	44.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0540	NM_015213	Huesken	RAB6IP1	23258	GGCCCGCAGCCUCUUCAGCuu	GCUGAAGAGGCUGCGGGCC	-3.4	-3.3	-3.1	2	1	7	-47.5	36.6	12.8	-2	II	73.7	31.4	37.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0541	NM_015213	Huesken	RAB6IP1	23258	UUCAGCUUGGAGGCACUCUca	AGAGUGCCUCCAAGCUGAA	-2.1	-0.9	-0.4	2	1	3	-41.4	65.5	-1.7	7	II	52.6	64.2	66.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0542	NM_015213	Huesken	RAB6IP1	23258	GUUGGGGAACUGUGGCAAGuc	CUUGCCACAGUUCCCCAAC	-2.1	-2.2	2.2	-1	1	4	-40.9	38.7	0	2	II	57.9	38.8	50.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0543	NM_015213	Huesken	RAB6IP1	23258	GGCAAGUCCUCUGGCAACUca	AGUUGCCAGAGGACUUGCC	-2.1	-3.3	0.4	1	-2	3	-42.4	43.3	-0.6	2	III	57.9	34.6	71.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0544	NM_015213	Huesken	RAB6IP1	23258	CACAAAGCAGAGGUUAGCCuc	GGCUAACCUCUGCUUUGUG	-3.3	-2.1	-1.5	-1	2	3	-38.9	56.3	12.7	3	II	52.6	52.3	70.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0545	NM_015213	Huesken	RAB6IP1	23258	CAGAGGUUAGCCUCUUGAGgc	CUCAAGAGGCUAACCUCUG	-2.1	-2.1	-5.9	0	0	3	-39.3	60.6	-6.7	2	II	52.6	47.4	52.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0546	NM_015213	Huesken	RAB6IP1	23258	AGUCAUCAGUCUCUGGUAAug	UUACCAGAGACUGAUGACU	-0.9	-2.1	0.1	3	0	2	-37.5	64.2	4.4	5	II	42.1	36.2	28.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0547	NM_015213	Huesken	RAB6IP1	23258	UGGCCCAUAGACCCCAGAAaa	UUCUGGGGUCUAUGGGCCA	-0.9	-2.1	-5.2	1	-1	5	-44	54.6	-1.4	2	II	57.9	45.7	63.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0548	NM_015213	Huesken	RAB6IP1	23258	AGGAGGUGGGAGCGGCACCuc	GGUGCCGCUCCCACCUCCU	-3.3	-2.1	-3.6	2	2	5	-48.7	34.5	12.1	2	II	73.7	52.4	39	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0549	NM_015213	Huesken	RAB6IP1	23258	UGCCUGGUGGAGUUGCUCCag	GGAGCAACUCCACCAGGCA	-3.3	-2.1	-3.8	2	3	3	-44.8	62.7	12	4	II	63.2	65.4	62.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0550	NM_015213	Huesken	RAB6IP1	23258	UCCAGCACGCUCCGACAUGcc	CAUGUCGGAGCGUGCUGGA	-2.1	-2.4	-2.4	0	1	3	-43.5	41.6	0.1	3	II	63.2	53.3	60.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0551	NM_015213	Huesken	RAB6IP1	23258	UGCACUUAGAGACGUAGAGag	CUCUACGUCUCUAAGUGCA	-2.1	-2.1	0.1	1	2	2	-37.8	69.1	2.4	6	Ib	47.4	57.9	68.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0552	NM_015213	Huesken	RAB6IP1	23258	UGUGAGGGCAAACCCAAAUgu	AUUUGGGUUUGCCCUCACA	-1.1	-2.1	-4.4	1	0	4	-38.6	58.9	3.4	6	II	47.4	56.3	72.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0553	NM_015213	Huesken	RAB6IP1	23258	UGUGAUAAUAAAGGCAUGGaa	CCAUGCCUUUAUUAUCACA	-3.3	-2.1	3.4	-1	3	3	-33.7	71	5	8	Ia	36.8	71	72.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0554	NM_015213	Huesken	RAB6IP1	23258	AAGGCAUGGAAUUGGGGCUcc	AGCCCCAAUUCCAUGCCUU	-2.1	-0.9	1.4	2	2	5	-41.1	60.4	1.4	5	II	52.6	53.3	57.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0555	NM_015213	Huesken	RAB6IP1	23258	CUGGGAUCAGCCUGGGUCUug	AGACCCAGGCUGAUCCCAG	-2.1	-2.1	-0.5	-1	-2	3	-45.1	41.6	-10.7	2	III	63.2	41.8	75.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0556	NM_015213	Huesken	RAB6IP1	23258	CCUACUGCAUCUUGGUCAAag	UUGACCAAGAUGCAGUAGG	-0.9	-3.3	1.3	-2	-3	2	-38.1	58.4	-5.1	3	III	47.4	26.4	57.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0557	NM_015213	Huesken	RAB6IP1	23258	AAAAUUUUCUCCUUCAGUCgu	GACUGAAGGAGAAAAUUUU	-2.4	-0.9	1.2	3	5	2	-30.9	96.7	5	7	Ia	31.6	75.7	90.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0558	NM_015213	Huesken	RAB6IP1	23258	AAGGGCUGGCACCAUCCCUgg	AGGGAUGGUGCCAGCCCUU	-2.1	-0.9	-3.1	3	2	4	-45.8	45.6	-1.6	2	II	63.2	54.1	83.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0559	NM_015213	Huesken	RAB6IP1	23258	CCAUCCCUGGCUUUAGAAGcc	CUUCUAAAGCCAGGGAUGG	-2.1	-3.3	-1.1	0	0	3	-39.1	62.2	1	1	II	52.6	37.6	46.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0560	NM_015213	Huesken	RAB6IP1	23258	ACUGGCAUAAUGCCGACAGcu	CUGUCGGCAUUAUGCCAGU	-2.1	-2.2	-4.9	1	3	4	-39.6	57.1	2.4	4	II	52.6	54.7	73	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0561	NM_015213	Huesken	RAB6IP1	23258	GUGAUGUUGAACUCCUGCAau	UGCAGGAGUUCAACAUCAC	-2.1	-2.2	0.4	2	-1	2	-38.1	69.9	-3.8	5	II	47.4	49.8	62.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0562	NM_015213	Huesken	RAB6IP1	23258	AAUGUCUGCAGCACACGAAuc	UUCGUGUGCUGCAGACAUU	-0.9	-0.9	-0.4	3	1	2	-38.2	62.8	-8.7	7	II	47.4	50.7	40.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0563	NM_015213	Huesken	RAB6IP1	23258	GAGGUGAUCUCUGGCUCCCaa	GGGAGCCAGAGAUCACCUC	-3.3	-2.4	-0.6	0	2	3	-44.5	44.8	8.1	2	II	63.2	48.8	49.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0564	NM_015213	Huesken	RAB6IP1	23258	AUCGGCAGAAGUUCCGGGCuc	GCCCGGAACUUCUGCCGAU	-3.4	-1.1	-2.8	1	4	6	-43.4	58.1	10.4	3	II	63.2	61	70.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0565	NM_015213	Huesken	RAB6IP1	23258	AGUGCUUCACAAUGCCAUUga	AAUGGCAUUGUGAAGCACU	-0.9	-2.1	-1	1	0	3	-36.5	67.5	-1.7	3	II	42.1	49.1	60.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0566	NM_015213	Huesken	RAB6IP1	23258	GACUGCUGCAGCGGCGGGGuc	CCCCGCCGCUGCAGCAGUC	-3.3	-2.4	-1.5	1	2	9	-48.8	33.1	-2.4	2	II	78.9	39.2	61.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0567	NM_015213	Huesken	RAB6IP1	23258	UCCGGCAUGGCCUCUCAUCca	GAUGAGAGGCCAUGCCGGA	-2.4	-2.4	-2.9	1	1	5	-44.7	52.3	2.7	2	II	63.2	58.7	63	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0568	NM_015213	Huesken	RAB6IP1	23258	AUGGCCUCUCAUCCACCUCag	GAGGUGGAUGAGAGGCCAU	-2.4	-1.1	-2.4	2	3	4	-43.2	64	15.1	2	II	57.9	61.5	74.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0569	NM_015213	Huesken	RAB6IP1	23258	CAGCUCCCCAACUAGGAUCcg	GAUCCUAGUUGGGGAGCUG	-2.4	-2.1	-3.4	1	-1	4	-41.9	59.4	5.4	1	II	57.9	50.9	66.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0570	NM_015213	Huesken	RAB6IP1	23258	GCUUCCAUCAUCCAUGCCCuu	GGGCAUGGAUGAUGGAAGC	-3.3	-3.4	0.2	1	2	4	-41.9	62.1	7.5	3	II	57.9	52.6	54.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0571	NM_015213	Huesken	RAB6IP1	23258	UCCACCAGCCAUUUGGCAUac	AUGCCAAAUGGCUGGUGGA	-1.1	-2.4	-4.1	1	1	3	-41.2	72.3	3.7	4	II	52.6	50.9	69.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0572	NM_015213	Huesken	RAB6IP1	23258	AUCCAUGGGUUGGCAGUGAac	UCACUGCCAACCCAUGGAU	-2.4	-1.1	-1.4	3	0	3	-41.5	43.3	-1.1	5	II	52.6	44.4	57.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0573	NM_015213	Huesken	RAB6IP1	23258	AAUCGACGGCAUUGAAAGAca	UCUUUCAAUGCCGUCGAUU	-2.4	-0.9	2.2	2	1	4	-35.4	55.3	1.6	4	II	42.1	40.9	52.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0574	NM_015213	Huesken	RAB6IP1	23258	GAAAGACAGGAGGUGAUAGag	CUAUCACCUCCUGUCUUUC	-2.1	-2.4	2.6	1	2	2	-37.4	53.9	4.6	4	II	47.4	42.2	63.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0575	NM_015213	Huesken	RAB6IP1	23258	GGAUGUGCCUCAUAUCCUGaa	CAGGAUAUGAGGCACAUCC	-2.1	-3.3	-0.9	1	1	3	-39.9	52.1	-2.4	3	II	52.6	44.2	46.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0576	NM_015213	Huesken	RAB6IP1	23258	GGAGGCAUGAGUGAGCUGGca	CCAGCUCACUCAUGCCUCC	-3.3	-3.3	-0.8	1	1	3	-44	44	3	3	II	63.2	45.3	64.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0577	NM_015213	Huesken	RAB6IP1	23258	CCUACGUUCUGAAUCAAGAag	UCUUGAUUCAGAACGUAGG	-2.4	-3.3	0.5	1	-2	2	-35.3	66	-6	3	III	42.1	42.4	29.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0578	NM_015213	Huesken	RAB6IP1	23258	UUUCUCUGCCGGUUGUCCUga	AGGACAACCGGCAGAGAAA	-2.1	-0.9	-0.7	3	4	5	-40.5	79.2	-1.6	7	II	52.6	67.4	67	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0579	NM_015213	Huesken	RAB6IP1	23258	AACAGGUGGGACCAUAAGGcu	CCUUAUGGUCCCACCUGUU	-3.3	-0.9	-0.8	3	3	3	-40.2	58.9	2.4	5	Ib	52.6	62.3	62.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0580	NM_015213	Huesken	RAB6IP1	23258	GAUCCUUUCCAGGAGAUCAca	UGAUCUCCUGGAAAGGAUC	-2.1	-2.4	-5.3	3	-1	2	-38.8	66.5	-1.5	3	II	47.4	45.6	44.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0581	NM_015213	Huesken	RAB6IP1	23258	UUUCCAGGAGAUCACAAAGgc	CUUUGUGAUCUCCUGGAAA	-2.1	-0.9	0.8	1	3	2	-35.7	73.5	7.7	6	Ib	42.1	65.6	65.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0582	NM_015213	Huesken	RAB6IP1	23258	AGGAGAUCACAAAGGCUGGca	CCAGCCUUUGUGAUCUCCU	-3.3	-2.1	1.8	1	2	3	-40.3	50.6	-0.3	5	Ib	52.6	52.8	47.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0583	NM_015213	Huesken	RAB6IP1	23258	GAUCACAAAGGCUGGCAAUca	AUUGCCAGCCUUUGUGAUC	-1.1	-2.4	0.6	0	-1	3	-37.9	53	-3.6	4	II	47.4	33.7	48.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0584	NM_015213	Huesken	RAB6IP1	23258	UCUUGGUCUUAUUGCGGCAuu	UGCCGCAAUAAGACCAAGA	-2.1	-2.4	3.3	0	1	5	-38.4	67.6	1.6	6	II	47.4	53	69.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0585	NM_015213	Huesken	RAB6IP1	23258	GAUGUACUUCUCCCUCUGGuc	CCAGAGGGAGAAGUACAUC	-3.3	-2.4	-0.2	0	2	3	-39.8	58.7	0.1	4	II	52.6	46.6	34.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0586	NM_015213	Huesken	RAB6IP1	23258	CUCUGGUCAUUAUCUAAACgc	GUUUAGAUAAUGACCAGAG	-2.2	-2.1	0.5	0	-1	2	-32.9	73.4	5.4	3	II	36.8	51.6	51.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0587	NM_015213	Huesken	RAB6IP1	23258	CAGUGGAAAGUACAUCAGAcu	UCUGAUGUACUUUCCACUG	-2.4	-2.1	1.6	-1	-3	2	-36	50.4	-3.4	3	III	42.1	43.5	47.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0588	NM_015213	Huesken	RAB6IP1	23258	AUUCAACAGCCUGAUCUUGuc	CAAGAUCAGGCUGUUGAAU	-2.1	-1.1	-0.4	3	3	3	-35.7	60	3.4	6	Ia	42.1	55	55.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0589	NM_015213	Huesken	RAB6IP1	23258	GGUCUCCAGGAAUCUUGAGag	CUCAAGAUUCCUGGAGACC	-2.1	-3.3	-0.5	2	1	2	-39.6	54.2	2.3	2	II	52.6	32.6	64.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0590	NM_015213	Huesken	RAB6IP1	23258	GCUUUAUCAAAGUUUUGCAuu	UGCAAAACUUUGAUAAAGC	-2.1	-3.4	1	1	0	2	-30.6	85.3	0.9	5	II	31.6	41.8	46	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0591	NM_015213	Huesken	RAB6IP1	23258	GGAUGACAAACGCCUCAUAau	UAUGAGGCGUUUGUCAUCC	-1.3	-3.3	-1.4	0	-3	4	-37.7	47.1	-3.7	2	III	47.4	26.7	44.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0592	NM_015213	Huesken	RAB6IP1	23258	GCCUCAUAAUCUGCAAACAuc	UGUUUGCAGAUUAUGAGGC	-2.1	-3.4	1	1	-3	3	-35.9	57.4	-3.1	2	III	42.1	29.7	38.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0593	NM_015213	Huesken	RAB6IP1	23258	UGCAAACAUCUGAGUGAAAcg	UUUCACUCAGAUGUUUGCA	-0.9	-2.1	1.3	-1	-2	2	-34.4	53.5	-3.6	5	II	36.8	39.5	42.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0594	NM_015213	Huesken	RAB6IP1	23258	AACUUCCCGGAUCUGAAUGuu	CAUUCAGAUCCGGGAAGUU	-2.1	-0.9	0.3	2	3	5	-37.4	77.7	10.1	5	Ib	47.4	57.2	74.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0595	NM_015213	Huesken	RAB6IP1	23258	UUCUUCUUCAUCACACUGAac	UCAGUGUGAUGAAGAAGAA	-2.4	-0.9	-0.5	2	0	1	-34.9	86.1	-8.7	9	II	36.8	68.8	75.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0596	NM_015213	Huesken	RAB6IP1	23258	ACUUUGAGAUCCUUAUUGCug	GCAAUAAGGAUCUCAAAGU	-3.4	-2.2	2.5	1	4	2	-33.7	81.5	5	7	Ia	36.8	60.5	66.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0597	NM_015213	Huesken	RAB6IP1	23258	UCCUUAUUGCUGCUGGGGUcu	ACCCCAGCAGCAAUAAGGA	-2.2	-2.4	0.4	2	2	4	-41.4	65.5	0.8	6	II	52.6	53.3	64.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0598	NM_015213	Huesken	RAB6IP1	23258	ACGCACUUCCAACUUUUCCag	GGAAAAGUUGGAAGUGCGU	-3.3	-2.2	0.7	4	3	3	-36.7	73.5	17.1	5	II	47.4	61.6	60.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0599	NM_015213	Huesken	RAB6IP1	23258	CUUAAGAAGCUCGUAGGAAug	UUCCUACGAGCUUCUUAAG	-0.9	-2.1	1.7	0	-1	2	-35.2	51	-3.1	5	II	42.1	39.2	59.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0600	NM_015213	Huesken	RAB6IP1	23258	UUAAGAAGCUCGUAGGAAUgc	AUUCCUACGAGCUUCUUAA	-1.1	-0.9	1.4	1	1	2	-34.2	69.5	-6.3	7	II	36.8	60.9	65.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0601	NM_015213	Huesken	RAB6IP1	23258	GAAGCUCGUAGGAAUGCAAag	UUGCAUUCCUACGAGCUUC	-0.9	-2.4	0.4	0	-1	2	-37.5	51.2	-6.3	3	III	47.4	35.1	50.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0602	NM_015213	Huesken	RAB6IP1	23258	AGCCAGCAAUGUUCCCAUUcc	AAUGGGAACAUUGCUGGCU	-0.9	-2.1	-2.6	2	-1	3	-38.8	55.8	1.8	2	II	47.4	39.1	43.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0603	NM_015213	Huesken	RAB6IP1	23258	GCCAGCAAUGUUCCCAUUCcu	GAAUGGGAACAUUGCUGGC	-2.4	-3.4	-3	0	-1	3	-39.1	56.4	12.8	2	II	52.6	45.6	45.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0604	NM_015213	Huesken	RAB6IP1	23258	GCAAUGUUCCCAUUCCUCUug	AGAGGAAUGGGAACAUUGC	-2.1	-3.4	-3	2	0	3	-38.1	67.2	-4	4	II	47.4	49.2	53.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0605	NM_015213	Huesken	RAB6IP1	23258	CAAUGUUCCCAUUCCUCUUgu	AAGAGGAAUGGGAACAUUG	-0.9	-2.1	-3	1	-1	3	-35.6	79	-0.7	3	II	42.1	52.1	50	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0606	NM_015213	Huesken	RAB6IP1	23258	GCCCGCAGCCUCUUCAGCUug	AGCUGAAGAGGCUGCGGGC	-2.1	-3.4	-3	2	0	6	-46.3	37.3	-4	-1	III	68.4	35.3	54.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0607	NM_015213	Huesken	RAB6IP1	23258	CCAAUUUGUUGGGGAACUGug	CAGUUCCCCAACAAAUUGG	-2.1	-3.3	-1.1	-1	0	4	-36.1	52.9	-2	3	II	47.4	45.1	67.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0608	NM_015213	Huesken	RAB6IP1	23258	CAAUGAAGUGGUUGUCAAUgu	AUUGACAACCACUUCAUUG	-1.1	-2.1	0.6	-2	-2	2	-33.1	64.1	-3	4	II	36.8	39.3	48.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0609	NM_015213	Huesken	RAB6IP1	23258	AGCUUUGACCGGUCAUCCAgg	UGGAUGACCGGUCAAAGCU	-2.1	-2.1	-1	2	0	4	-40.7	61.6	-3.8	4	II	52.6	45.5	67.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0610	NM_015213	Huesken	RAB6IP1	23258	AUCUAAGAAAUGCAGGAGAga	UCUCCUGCAUUUCUUAGAU	-2.4	-1.1	1.2	1	0	2	-35.1	63.1	1.4	7	II	36.8	51.9	51.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0611	NM_015213	Huesken	RAB6IP1	23258	UUUGACAGGAAAGUCAAAUag	AUUUGACUUUCCUGUCAAA	-1.1	-0.9	-4.9	1	2	2	-31.9	77.3	3.4	7	II	31.6	66.6	53.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0612	NM_015213	Huesken	RAB6IP1	23258	UAUUGGCCCAUAGACCCCAga	UGGGGUCUAUGGGCCAAUA	-2.1	-1.3	-5.2	1	2	5	-41.9	54.1	-3.8	6	II	52.6	63.2	66.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0613	NM_015213	Huesken	RAB6IP1	23258	CCAGGAGGUGGGAGCGGCAcc	UGCCGCUCCCACCUCCUGG	-2.1	-3.3	-0.3	-2	-2	5	-48.6	10.4	-8.3	1	III	73.7	27.2	24.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0614	NM_015213	Huesken	RAB6IP1	23258	UGUAUAUGUAGCUCUCAAGgg	CUUGAGAGCUACAUAUACA	-2.1	-2.1	2.4	1	2	2	-34.3	87.4	-2.4	8	Ia	36.8	66.6	74.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0615	NM_015213	Huesken	RAB6IP1	23258	CAUAGGAGUUGAAGCGCUGca	CAGCGCUUCAACUCCUAUG	-2.1	-2.1	2.5	-2	1	4	-38.4	44.6	-4.6	4	II	52.6	48.3	49.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0616	NM_015213	Huesken	RAB6IP1	23258	UGCAGUUUGGUCACAGGAGug	CUCCUGUGACCAAACUGCA	-2.1	-2.1	0.6	2	2	2	-40	59.6	-4.9	5	Ib	52.6	60.5	80.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0617	NM_015213	Huesken	RAB6IP1	23258	CAGGAGUGUCUUCACCAUCcu	GAUGGUGAAGACACUCCUG	-2.4	-2.1	-1.7	0	-1	2	-39.5	52.2	13.5	2	II	52.6	54.9	82.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0618	NM_015213	Huesken	RAB6IP1	23258	ACUCAGCAUUGUGCAUGUGgu	CACAUGCACAAUGCUGAGU	-2.1	-2.2	-0.8	2	3	2	-37.8	51.6	5.7	4	II	47.4	44.6	69.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0619	NM_015213	Huesken	RAB6IP1	23258	AUUGUGCAUGUGGUAGAGGgu	CCUCUACCACAUGCACAAU	-3.3	-1.1	3.3	1	5	2	-38.1	57.4	2.3	5	Ib	47.4	61.1	68.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0620	NM_015213	Huesken	RAB6IP1	23258	GCAUUGCACUGCAGAUCUGcu	CAGAUCUGCAGUGCAAUGC	-2.1	-3.4	-1.2	0	1	2	-39.2	43.5	-4.9	2	II	52.6	34.9	56.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0621	NM_015213	Huesken	RAB6IP1	23258	CUUGCUAGUCACCUCUUCAua	UGAAGAGGUGACUAGCAAG	-2.1	-2.1	-2.5	-1	-1	2	-38.2	70	-1.1	3	II	47.4	49.6	68.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0622	NM_015213	Huesken	RAB6IP1	23258	AUGUGAGGGCAAACCCAAAug	UUUGGGUUUGCCCUCACAU	-0.9	-1.1	-4.4	2	-1	4	-38.6	47.2	-4	4	II	47.4	43.6	73.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0623	NM_015213	Huesken	RAB6IP1	23258	AUGUCCGAGAGCCAUCCUCcc	GAGGAUGGCUCUCGGACAU	-2.4	-1.1	-2.6	2	3	3	-42.6	60.3	7.5	4	II	57.9	61.2	55.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0624	NM_015213	Huesken	RAB6IP1	23258	GGCUCCCUGGGAUCAGCCUgg	AGGCUGAUCCCAGGGAGCC	-2.1	-3.3	-3.2	1	-1	3	-47.5	42.3	-4	-1	III	68.4	27.9	57.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0625	NM_015213	Huesken	RAB6IP1	23258	UGGGUCUUGAAUGCCAGCCcu	GGCUGGCAUUCAAGACCCA	-3.3	-2.1	0.8	2	3	3	-42.5	68.8	12.7	5	II	57.9	71.7	75.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0626	NM_015213	Huesken	RAB6IP1	23258	AUUUGAAUGUUCUUCUCAAug	UUGAGAAGAACAUUCAAAU	-0.9	-1.1	-0.7	2	2	1	-29.9	82.7	-3.8	6	II	26.3	52.9	54.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0627	NM_015213	Huesken	RAB6IP1	23258	CAUAAUGCCGACAGCUCGUcc	ACGAGCUGUCGGCAUUAUG	-2.2	-2.1	-1.5	-1	0	4	-39	53.5	-5.9	4	II	52.6	47.7	54.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0628	NM_015213	Huesken	RAB6IP1	23258	CGGUCUCCGUGUCCAGUCCgc	GGACUGGACACGGAGACCG	-3.3	-2.4	-1.9	0	-1	3	-44.8	52.6	10.8	0	II	68.4	40.2	60.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0629	NM_003347	Huesken	UBE2L3	7332	AGAGUAUUCUUCAGCUAGGuc	CCUAGCUGAAGAAUACUCU	-3.3	-2.1	0.8	2	3	2	-36.3	61.6	-4.9	5	Ia	42.1	59.5	63.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0630	NM_003347	Huesken	UBE2L3	7332	AGUAUUCUUCAGCUAGGUCag	GACCUAGCUGAAGAAUACU	-2.4	-2.1	1.4	1	4	2	-36.4	77.8	14.4	5	Ia	42.1	56.7	101.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0631	NM_003347	Huesken	UBE2L3	7332	AUUCUUCAGCUAGGUCAGCcc	GCUGACCUAGCUGAAGAAU	-3.4	-1.1	-1.5	2	4	2	-38.4	54.9	7.4	4	Ia	47.4	58.1	70.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0632	NM_003347	Huesken	UBE2L3	7332	CUUCAGCUAGGUCAGCCCGaa	CGGGCUGACCUAGCUGAAG	-2.4	-2.1	-3	0	2	5	-43	49	6.1	1	II	63.2	50.7	61.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0633	NM_003347	Huesken	UBE2L3	7332	UCAGCUAGGUCAGCCCGAAgc	UUCGGGCUGACCUAGCUGA	-0.9	-2.4	-3	1	0	5	-43.3	39.4	-6.1	3	II	57.9	44.8	49	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0634	NM_003347	Huesken	UBE2L3	7332	UAGGUCAGCCCGAAGCGGGug	CCCGCUUCGGGCUGACCUA	-3.3	-1.3	-3.7	1	4	5	-45.7	45.6	-4.9	4	Ib	68.4	66.6	62.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0635	NM_003347	Huesken	UBE2L3	7332	GGUCAGCCCGAAGCGGGUGcu	CACCCGCUUCGGGCUGACC	-2.1	-3.3	-4.2	2	1	5	-46.6	25.4	5	0	II	73.7	31.9	60	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0636	NM_003347	Huesken	UBE2L3	7332	UCAGCCCGAAGCGGGUGCUca	AGCACCCGCUUCGGGCUGA	-2.1	-2.4	-4.2	-1	2	5	-46.6	37.9	-6.3	3	II	68.4	46.3	37.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0637	NM_003347	Huesken	UBE2L3	7332	GCCCGAAGCGGGUGCUCAGgc	CUGAGCACCCGCUUCGGGC	-2.1	-3.4	-4.2	1	0	5	-46.6	24.1	0	-1	II	73.7	25.7	51.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0638	NM_003347	Huesken	UBE2L3	7332	CGAAGCGGGUGCUCAGGCUgg	AGCCUGAGCACCCGCUUCG	-2.1	-2.4	-2.5	0	-1	5	-45.4	34.1	-8.3	1	III	68.4	34.5	46.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0639	NM_003347	Huesken	UBE2L3	7332	GAAGCGGGUGCUCAGGCUGgg	CAGCCUGAGCACCCGCUUC	-2.1	-2.4	-2.5	0	2	5	-45.1	36.2	3.1	1	II	68.4	45.1	37.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0640	NM_003347	Huesken	UBE2L3	7332	GUCAUUCACCAGUGCUAUGag	CAUAGCACUGGUGAAUGAC	-2.1	-2.2	0.7	0	0	2	-37.2	65.6	2.3	3	II	47.4	44.1	80.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0641	NM_003347	Huesken	UBE2L3	7332	UCAUUCACCAGUGCUAUGAgg	UCAUAGCACUGGUGAAUGA	-2.4	-2.4	-0.9	1	0	2	-37.4	79	-0.7	9	II	42.1	60.4	77.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0642	NM_003347	Huesken	UBE2L3	7332	CAUUCACCAGUGCUAUGAGgg	CUCAUAGCACUGGUGAAUG	-2.1	-2.1	-1	-1	2	2	-37.1	63.4	2.7	2	II	47.4	40.9	77.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0643	NM_003347	Huesken	UBE2L3	7332	GAGGGACUGGAUUACUUGGuc	CCAAGUAAUCCAGUCCCUC	-3.3	-2.4	3.6	1	1	3	-39.5	56.4	5.3	2	II	52.6	45.3	61.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0644	NM_003347	Huesken	UBE2L3	7332	AGGGACUGGAUUACUUGGUcg	ACCAAGUAAUCCAGUCCCU	-2.2	-2.1	3.3	3	1	3	-39.3	58.7	-0.8	5	II	47.4	51.7	87.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0645	NM_003347	Huesken	UBE2L3	7332	CUGGAUUACUUGGUCGGUUuu	AACCGACCAAGUAAUCCAG	-0.9	-2.1	2	1	-1	3	-37.2	54.8	-0.9	3	II	47.4	46.5	66.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0646	NM_003347	Huesken	UBE2L3	7332	GAUUACUUGGUCGGUUUUGgu	CAAAACCGACCAAGUAAUC	-2.1	-2.4	2.2	1	1	3	-33.6	67.1	0.4	6	II	42.1	47.4	56.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0647	NM_003347	Huesken	UBE2L3	7332	AUUACUUGGUCGGUUUUGGuu	CCAAAACCGACCAAGUAAU	-3.3	-1.1	2.2	3	5	3	-34.5	81.4	0	7	Ia	42.1	64.6	84.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0648	NM_003347	Huesken	UBE2L3	7332	ACUUGGUCGGUUUUGGUUGcu	CAACCAAAACCGACCAAGU	-2.1	-2.2	2.9	3	3	3	-36.4	73.6	5.4	6	Ib	47.4	59.1	87	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0649	NM_003347	Huesken	UBE2L3	7332	UGGUCGGUUUUGGUUGCUGgc	CAGCAACCAAAACCGACCA	-2.1	-2.1		0	3	3	-38.8	61.4	2.3	5	II	52.6	59	60	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0650	NM_003347	Huesken	UBE2L3	7332	CGGUUUUGGUUGCUGGCUUcc	AAGCCAGCAACCAAAACCG	-0.9	-2.4		1	-2	3	-38.5	58.4	-1.2	3	II	52.6	38.9	41.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0651	NM_003347	Huesken	UBE2L3	7332	GUUUUGGUUGCUGGCUUCCag	GGAAGCCAGCAACCAAAAC	-3.3	-2.2	0.4	-1	2	3	-38.5	62.8	8.1	6	II	52.6	52.9	73	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0652	NM_003347	Huesken	UBE2L3	7332	UGGUUGCUGGCUUCCAGUUuu	AACUGGAAGCCAGCAACCA	-0.9	-2.1	-2.7	1	0	3	-40.9	65	-3.3	4	II	52.6	52.7	71.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0653	NM_003347	Huesken	UBE2L3	7332	UUUCGGCACUAAUUACUGGca	CCAGUAAUUAGUGCCGAAA	-3.3	-0.9	0.8	2	5	4	-34.9	73.7	4.6	6	II	42.1	68.3	82.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0654	NM_003347	Huesken	UBE2L3	7332	CACUAAUUACUGGCAGACAga	UGUCUGCCAGUAAUUAGUG	-2.1	-2.1	1.6	0	-3	3	-35.9	57.7	-3.1	4	II	42.1	40.7	51.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0655	NM_003347	Huesken	UBE2L3	7332	ACUAAUUACUGGCAGACAGac	CUGUCUGCCAGUAAUUAGU	-2.1	-2.2	-0.2	3	3	3	-35.9	70.5	5.1	6	Ia	42.1	56.8	52.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0656	NM_003347	Huesken	UBE2L3	7332	UAAUUACUGGCAGACAGACcu	GUCUGUCUGCCAGUAAUUA	-2.2	-1.3	-0.3	1	4	3	-36.2	65.6	5.1	7	Ia	42.1	65.8	91	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0657	NM_003347	Huesken	UBE2L3	7332	AAUUACUGGCAGACAGACCug	GGUCUGUCUGCCAGUAAUU	-3.3	-0.9	-2.1	3	4	3	-38.2	70	9.8	6	Ia	47.4	66.4	74.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0658	NM_003347	Huesken	UBE2L3	7332	UACUGGCAGACAGACCUGCcc	GCAGGUCUGUCUGCCAGUA	-3.4	-1.3	-4.7	1	3	3	-42.9	52	5.1	5	II	57.9	66	52.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0659	NM_003347	Huesken	UBE2L3	7332	CCUGCCCCUUUUCGUCGAUgu	AUCGACGAAAAGGGGCAGG	-1.1	-3.3	1.3	-2	-2	5	-40.9	41.9	-0.6	0	III	57.9	22	32	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0660	NM_003347	Huesken	UBE2L3	7332	UGCCCCUUUUCGUCGAUGUuu	ACAUCGACGAAAAGGGGCA	-2.2	-2.1	1.3	2	0	5	-39.8	70.8	-6.3	5	II	52.6	53.7	64.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0661	NM_003347	Huesken	UBE2L3	7332	CCUUUUCGUCGAUGUUUGGgu	CCAAACAUCGACGAAAAGG	-3.3	-3.3	2	-2	0	2	-34.9	65.3	-2	4	II	47.4	40.7	59.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0662	NM_003347	Huesken	UBE2L3	7332	CGUCGAUGUUUGGGUGAUAga	UAUCACCCAAACAUCGACG	-1.3	-2.4	2.4	0	-4	3	-36.8	37.7	-5.8	2	III	47.4	25.2	57	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0663	NM_003347	Huesken	UBE2L3	7332	AUGUUUGGGUGAUAGAUCUuu	AGAUCUAUCACCCAAACAU	-2.1	-1.1	1	2	1	3	-35.1	77.5	-4	7	II	36.8	61.9	73.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0664	NM_003347	Huesken	UBE2L3	7332	UUUGGGUGAUAGAUCUUUGuu	CAAAGAUCUAUCACCCAAA	-2.1	-0.9	1	1	4	3	-33.6	81.2	2	7	Ib	36.8	72.9	82.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0665	NM_003347	Huesken	UBE2L3	7332	GGGUGAUAGAUCUUUGUUUua	AAACAAAGAUCUAUCACCC	-0.9	-3.3	1	2	-2	3	-33.7	70.4	-4	4	III	36.8	40.5	50.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0666	NM_003347	Huesken	UBE2L3	7332	AUAGAUCUUUGUUUUAAAUgu	AUUUAAAACAAAGAUCUAU	-1.1	-1.1	2.8	1	2	1	-25.5	97.7	4.1	5	II	15.8	57.9	65.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0667	NM_003347	Huesken	UBE2L3	7332	UUUUAAAUGUGAUCUUCGGug	CCGAAGAUCACAUUUAAAA	-3.3	-0.9	1.5	1	5	3	-30.3	89.1	2.7	8	Ia	31.6	72.4	83.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0668	NM_003347	Huesken	UBE2L3	7332	UGAUCUUCGGUGGUUUGAAug	UUCAAACCACCGAAGAUCA	-0.9	-2.1	0.5	2	1	3	-36.1	71.6	-1.4	6	II	42.1	51.9	65.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0669	NM_003347	Huesken	UBE2L3	7332	UCGGUGGUUUGAAUGGGUAcu	UACCCAUUCAAACCACCGA	-1.3	-2.4	2	0	0	3	-38.4	54.4	-4	5	II	47.4	49.5	71.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0670	NM_003347	Huesken	UBE2L3	7332	UUGAAUGGGUACUCUGCUGga	CAGCAGAGUACCCAUUCAA	-2.1	-0.9	-1	1	3	3	-38.1	71.5	2.7	6	Ia	47.4	69	90	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0671	NM_003347	Huesken	UBE2L3	7332	UCUGAAGGCUCCCUUAUCAua	UGAUAAGGGAGCCUUCAGA	-2.1	-2.4	-1	1	1	3	-39.7	60.5	-1	6	II	47.4	53.3	109.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0672	NM_003347	Huesken	UBE2L3	7332	CUCCCUUAUCAUAUGGAGGgu	CCUCCAUAUGAUAAGGGAG	-3.3	-2.1	-3	1	1	3	-37.7	73.1	3.1	1	II	47.4	55.3	81.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0673	NM_003347	Huesken	UBE2L3	7332	AUAUGGAGGGUUGUCAGGAac	UCCUGACAACCCUCCAUAU	-2.4	-1.1	2	2	3	3	-39.7	55.1	1.6	6	II	47.4	56.2	69.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0674	NM_003347	Huesken	UBE2L3	7332	GGAACAAUAAGCCCUUGCCaa	GGCAAGGGCUUAUUGUUCC	-3.3	-3.3	-1.9	0	2	4	-39	63.7	7.5	3	II	52.6	49.8	79.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0675	NM_003347	Huesken	UBE2L3	7332	UAAGCCCUUGCCAAGUCAAua	UUGACUUGGCAAGGGCUUA	-0.9	-1.3	-1.1	0	1	4	-38.8	59.2	-11	4	II	47.4	53.4	61.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0676	NM_003347	Huesken	UBE2L3	7332	AAGCCCUUGCCAAGUCAAUaa	AUUGACUUGGCAAGGGCUU	-1.1	-0.9	-1.1	3	0	4	-38.6	54.4	-8.9	3	II	47.4	44.7	58.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0677	NM_003347	Huesken	UBE2L3	7332	AGCCCUUGCCAAGUCAAUAaa	UAUUGACUUGGCAAGGGCU	-1.3	-2.1	-1	4	-1	4	-39	59.9	0.9	2	II	47.4	44.7	75.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0678	NM_003347	Huesken	UBE2L3	7332	GCCAAGUCAAUAAAUUAGCuu	GCUAAUUUAUUGACUUGGC	-3.4	-3.4	-1.3	1	0	3	-32.4	64	7.5	3	II	36.8	52.6	46.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0679	NM_003347	Huesken	UBE2L3	7332	AAUUAGCUUCAUCAACCUGga	CAGGUUGAUGAAGCUAAUU	-2.1	-0.9	-1.5	1	4	2	-33.4	75.8	10.4	5	Ia	36.8	61	96.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0680	NM_003347	Huesken	UBE2L3	7332	UUAGCUUCAUCAACCUGGAug	UCCAGGUUGAUGAAGCUAA	-2.4	-0.9	-1.5	1	2	2	-37.1	69.3	-8.7	5	II	42.1	63.7	111.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0681	NM_003347	Huesken	UBE2L3	7332	AUCAACCUGGAUGUUACGGaa	CCGUAACAUCCAGGUUGAU	-3.3	-1.1	-0.6	2	5	3	-37.4	70.9	10.4	5	Ib	47.4	60.2	61.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0682	NM_003344	Huesken	UBE2H	7328	UGAGCUGUCCCCGGUACCCuc	GGGUACCGGGGACAGCUCA	-3.3	-2.1	-1.8	0	4	6	-46.7	46	2.7	3	II	68.4	63.8	71.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0683	NM_003344	Huesken	UBE2H	7328	GAGCUGUCCCCGGUACCCUcu	AGGGUACCGGGGACAGCUC	-2.1	-2.4	-0.3	3	1	6	-46.7	38.4	-4	2	III	68.4	42	77.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0684	NM_003344	Huesken	UBE2H	7328	GCUGUCCCCGGUACCCUCUuc	AGAGGGUACCGGGGACAGC	-2.1	-3.4	0.5	1	-1	6	-46.7	44.5	-3.5	2	III	68.4	35.8	65	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0685	NM_003344	Huesken	UBE2H	7328	CCCCGGUACCCUCUUCCUGuu	CAGGAAGAGGGUACCGGGG	-2.1	-3.3	0.1	2	0	6	-45.3	41.1	3.4	0	II	68.4	39.3	74.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0686	NM_003344	Huesken	UBE2H	7328	UUCCUGUUCUUUCAGCGCCuc	GGCGCUGAAAGAACAGGAA	-3.3	-0.9	-1.4	3	4	5	-39.3	72.7	12.8	5	Ib	52.6	75.3	98.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0687	NM_003344	Huesken	UBE2H	7328	UUCUUUCAGCGCCUCCUCCgu	GGAGGAGGCGCUGAAAGAA	-3.3	-0.9	1.3	1	4	5	-41.9	75	9.8	6	Ia	57.9	70.9	91.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0688	NM_003344	Huesken	UBE2H	7328	CAGCGCCUCCUCCGUGGCGua	CGCCACGGAGGAGGCGCUG	-2.4	-2.1	-5.4	0	1	5	-48.1	27.2	-2	-1	II	78.9	38.9	66.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0689	NM_003344	Huesken	UBE2H	7328	AGCGCCUCCUCCGUGGCGUau	ACGCCACGGAGGAGGCGCU	-2.2	-2.1	-5.9	3	2	5	-48.2	40.6	-6.3	1	II	73.7	41.3	71.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0690	NM_003344	Huesken	UBE2H	7328	CGUAUUUCUGGAUGUACUCuu	GAGUACAUCCAGAAAUACG	-2.4	-2.4	1	-1	0	2	-34.5	73.3	5.4	4	II	42.1	52.8	77.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0691	NM_003344	Huesken	UBE2H	7328	GGAUGUACUCUUUAAUUUUcu	AAAAUUAAAGAGUACAUCC	-0.9	-3.3	1.1	0	-2	2	-29	78.4	-1	4	II	26.3	45.2	70.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0692	NM_003344	Huesken	UBE2H	7328	UGUACUCUUUAAUUUUCUGcu	CAGAAAAUUAAAGAGUACA	-2.1	-2.1	2.5	0	4	1	-28.8	99.9	4.6	5	Ib	26.3	66.6	78.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0693	NM_003344	Huesken	UBE2H	7328	UAAUUUUCUGCUUGUAUUCuu	GAAUACAAGCAGAAAAUUA	-2.4	-1.3	3.1	0	3	2	-29	103.8	8.1	9	Ia	26.3	79.7	93.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0694	NM_003344	Huesken	UBE2H	7328	UUUUCUGCUUGUAUUCUUCug	GAAGAAUACAAGCAGAAAA	-2.4	-0.9		0	4	2	-31.1	101.6	10.2	8	Ia	31.6	76	95.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0695	NM_003344	Huesken	UBE2H	7328	UUCUGCUUGUAUUCUUCUGgu	CAGAAGAAUACAAGCAGAA	-2.1	-0.9		2	3	2	-33.5	94.4	5.3	6	Ib	36.8	68.9	77.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0696	NM_003344	Huesken	UBE2H	7328	UCUGCUUGUAUUCUUCUGGuc	CCAGAAGAAUACAAGCAGA	-3.3	-2.4		1	4	2	-35.9	87.7	7.8	7	Ib	42.1	73.1	103.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0697	NM_003344	Huesken	UBE2H	7328	CUUGUAUUCUUCUGGUCGGug	CCGACCAGAAGAAUACAAG	-3.3	-2.1		0	2	3	-36.2	65.9	-4.4	4	II	47.4	53.6	86.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0698	NM_003344	Huesken	UBE2H	7328	GUAUUCUUCUGGUCGGUGGag	CCACCGACCAGAAGAAUAC	-3.3	-2.2		2	2	3	-38.7	76	0	6	II	52.6	53.4	82.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0699	NM_003344	Huesken	UBE2H	7328	UCUGGUCGGUGGAGGUACAug	UGUACCUCCACCGACCAGA	-2.1	-2.4	2.1	0	0	3	-43.4	41.6	-6.3	3	II	57.9	42.8	95.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0700	NM_003344	Huesken	UBE2H	7328	CGGUGGAGGUACAUGGCUGca	CAGCCAUGUACCUCCACCG	-2.1	-2.4	1.8	0	0	3	-43	40.8	-2.3	1	II	63.2	41.7	61.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0701	NM_003344	Huesken	UBE2H	7328	GGUACAUGGCUGCAGCGUCac	GACGCUGCAGCCAUGUACC	-2.4	-3.3	-1.2	2	1	3	-43.2	43.8	9.8	1	II	63.2	40.2	55.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0702	NM_003344	Huesken	UBE2H	7328	ACAUGGCUGCAGCGUCACCau	GGUGACGCUGCAGCCAUGU	-3.3	-2.2	-1.9	1	3	3	-44	43.3	12.1	2	II	63.2	49.4	91	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0703	NM_003344	Huesken	UBE2H	7328	AUGGCUGCAGCGUCACCAUug	AUGGUGACGCUGCAGCCAU	-1.1	-1.1	-2.9	1	1	3	-42.9	50.8	-6.3	2	II	57.9	48.2	86.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0704	NM_003344	Huesken	UBE2H	7328	UGAGAGGAUCUAUGGGGUUag	AACCCCAUAGAUCCUCUCA	-0.9	-2.1	1	-1	1	4	-39.9	45.5	-1.3	6	II	47.4	52.5	49.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0705	NM_003344	Huesken	UBE2H	7328	GAGGAUCUAUGGGGUUAGGau	CCUAACCCCAUAGAUCCUC	-3.3	-2.4	0.3	-1	1	4	-40	47.5	2.7	2	II	52.6	36.8	66.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0706	NM_003344	Huesken	UBE2H	7328	GAUCUAUGGGGUUAGGAUAgg	UAUCCUAACCCCAUAGAUC	-1.3	-2.4	0.3	3	-2	4	-37	64.3	-0.7	4	II	42.1	38	50.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0707	NM_003344	Huesken	UBE2H	7328	GGGGUUAGGAUAGGCCAAUaa	AUUGGCCUAUCCUAACCCC	-1.1	-3.3	0.4	0	-3	4	-40.7	36.2	-4	3	III	52.6	32.5	36.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0708	NM_003344	Huesken	UBE2H	7328	AGGAUAGGCCAAUAACUGAgg	UCAGUUAUUGGCCUAUCCU	-2.4	-2.1	0.6	2	-1	4	-37.4	58.4	-1.5	6	II	42.1	46.7	66.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0709	NM_003344	Huesken	UBE2H	7328	AGGCAGGAAGGACUCAAAUau	AUUUGAGUCCUUCCUGCCU	-1.1	-2.1	1.7	2	0	3	-39.1	49.1	3.5	4	II	47.4	43.2	48.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0710	NM_003344	Huesken	UBE2H	7328	GGAAGGACUCAAAUAUAUUgg	AAUAUAUUUGAGUCCUUCC	-0.9	-3.3	2.4	0	-1	2	-31.8	62.2	-1.9	3	III	31.6	42.8	41.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0711	NM_003344	Huesken	UBE2H	7328	GAAGGACUCAAAUAUAUUGgu	CAAUAUAUUUGAGUCCUUC	-2.1	-2.4	0.4	1	1	2	-30.6	73.2	2.3	4	II	31.6	54.1	84.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0712	NM_003344	Huesken	UBE2H	7328	GGACUCAAAUAUAUUGGUAag	UACCAAUAUAUUUGAGUCC	-1.3	-3.3	0.1	1	-1	2	-32	68.8	1.3	3	III	31.6	34.9	48.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0713	NM_003344	Huesken	UBE2H	7328	CAAAUAUAUUGGUAAGAUCau	GAUCUUACCAAUAUAUUUG	-2.4	-2.1	1.2	0	0	2	-28.5	81.3	5.4	5	II	26.3	58.7	78.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0714	NM_003344	Huesken	UBE2H	7328	AAAUAUAUUGGUAAGAUCAua	UGAUCUUACCAAUAUAUUU	-2.1	-0.9	2	1	1	2	-28.5	91.5	-0.6	8	II	21.1	68.6	68.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0715	NM_003344	Huesken	UBE2H	7328	UAUAUUGGUAAGAUCAUAGag	CUAUGAUCUUACCAAUAUA	-2.1	-1.3	1.7	0	4	2	-30.1	100	2	10	Ia	26.3	80	110	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0716	NM_003344	Huesken	UBE2H	7328	UAAGAUCAUAGAGAGCUGUcc	ACAGCUCUCUAUGAUCUUA	-2.2	-1.3	0.8	0	2	2	-35.6	68	-1.3	8	II	36.8	67.3	95.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0717	NM_003344	Huesken	UBE2H	7328	UCAUAGAGAGCUGUCCAAGuu	CUUGGACAGCUCUCUAUGA	-2.1	-2.4	-0.6	0	3	2	-38.6	62.3	5.7	7	Ia	47.4	61.7	81	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0718	NM_003344	Huesken	UBE2H	7328	CAUAGAGAGCUGUCCAAGUuu	ACUUGGACAGCUCUCUAUG	-2.2	-2.1	-0.6	0	-1	2	-38.4	54.2	-6	4	II	47.4	45.9	86	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0719	NM_003344	Huesken	UBE2H	7328	GUUUGAUUAAUUACAUCUAga	UAGAUGUAAUUAAUCAAAC	-1.3	-2.2	-0.9	1	-1	1	-27.2	83.7	-3.3	5	II	21.1	47.5	46.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0720	NM_003344	Huesken	UBE2H	7328	UGAUUAAUUACAUCUAGACac	GUCUAGAUGUAAUUAAUCA	-2.2	-2.1	-0.9	0	3	1	-29.9	91.9	5.1	8	Ia	26.3	71.3	105.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0721	NM_003344	Huesken	UBE2H	7328	ACAUCUAGACACACAGUUCcu	GAACUGUGUGUCUAGAUGU	-2.4	-2.2	0.4	1	2	1	-36.2	70.7	4.8	6	Ia	42.1	60.3	98.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0722	NM_003344	Huesken	UBE2H	7328	AUCUAGACACACAGUUCCUga	AGGAACUGUGUGUCUAGAU	-2.1	-1.1	0.3	0	2	2	-37.3	63.7	-3.9	6	II	42.1	55.7	100.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0723	NM_003344	Huesken	UBE2H	7328	AGACACACAGUUCCUGACGcu	CGUCAGGAACUGUGUGUCU	-2.4	-2.1	0.8	2	3	2	-39.5	58.7	8.1	5	Ib	52.6	60.2	99.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0724	NM_003344	Huesken	UBE2H	7328	AUGGAAAAUUUUAUUCAUGaa	CAUGAAUAAAAUUUUCCAU	-2.1	-1.1	0.7	1	3	2	-26.4	80.9	5.5	5	Ia	21.1	67.6	64	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0725	NM_003344	Huesken	UBE2H	7328	UUUUAUUCAUGAAUCCUAUag	AUAGGAUUCAUGAAUAAAA	-1.1	-0.9	1.8	1	2	2	-28.3	99.7	-1.5	9	II	21.1	69	106.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0726	NM_003344	Huesken	UBE2H	7328	UUUAUUCAUGAAUCCUAUAga	UAUAGGAUUCAUGAAUAAA	-1.3	-0.9	1.8	0	0	2	-28.7	95.4	-1.5	7	II	21.1	66.1	67.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0727	NM_003344	Huesken	UBE2H	7328	AUUCAUGAAUCCUAUAGAUgg	AUCUAUAGGAUUCAUGAAU	-1.1	-1.1	0.7	2	2	2	-31.2	76.7	-6	6	II	26.3	54.1	95.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0728	NM_003344	Huesken	UBE2H	7328	AUCCUAUAGAUGGAGAUUUga	AAAUCUCCAUCUAUAGGAU	-0.9	-1.1	-1.5	4	0	2	-33.4	70.7	-4	6	II	31.6	59.2	96.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0729	NM_003344	Huesken	UBE2H	7328	UGGAGAUUUGAAAGGGUAUuu	AUACCCUUUCAAAUCUCCA	-1.1	-2.1	5.1	0	-1	3	-34.8	65.8	-4.3	6	II	36.8	52.8	99.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0730	NM_003344	Huesken	UBE2H	7328	AUUUGAAAGGGUAUUUAUCag	GAUAAAUACCCUUUCAAAU	-2.4	-1.1	1.4	2	4	3	-29.1	88.6	10.2	6	Ia	26.3	67.1	101.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0731	NM_003344	Huesken	UBE2H	7328	AUUUAUCAGGUAGGUCCACuc	GUGGACCUACCUGAUAAAU	-2.2	-1.1	0	1	4	2	-36.3	58.6	10.1	6	Ia	42.1	55.6	77.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0732	NM_003344	Huesken	UBE2H	7328	UUUAUCAGGUAGGUCCACUcu	AGUGGACCUACCUGAUAAA	-2.1	-0.9	0	1	3	2	-37.3	77.5	0.7	8	II	42.1	69.1	108.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0733	NM_003344	Huesken	UBE2H	7328	GGUAGGUCCACUCUAACUUuc	AAGUUAGAGUGGACCUACC	-0.9	-3.3	0.3	2	0	2	-38.4	64.5	1.4	4	III	47.4	43.1	64.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0734	NM_003344	Huesken	UBE2H	7328	GUAGGUCCACUCUAACUUUcc	AAAGUUAGAGUGGACCUAC	-0.9	-2.2	0.3	0	-1	2	-36	56.5	-6.3	3	III	42.1	43.6	50	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0735	NM_003344	Huesken	UBE2H	7328	ACUCUAACUUUCCAUACUCcg	GAGUAUGGAAAGUUAGAGU	-2.4	-2.2	2.7	2	3	2	-34	67.7	7.5	5	Ia	36.8	60.2	83.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0736	NM_003344	Huesken	UBE2H	7328	AUACUCCGCCUUCAUAUGGug	CCAUAUGAAGGCGGAGUAU	-3.3	-1.1	0.6	3	4	5	-37.9	68.2	5.4	6	Ib	47.4	63.5	101.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0737	NM_003344	Huesken	UBE2H	7328	UCCGCCUUCAUAUGGUGUUcc	AACACCAUAUGAAGGCGGA	-0.9	-2.4	0.1	1	0	5	-38.6	66.2	-4	5	II	47.4	53.6	67.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0738	NM_003344	Huesken	UBE2H	7328	CUUCAUAUGGUGUUCCUUGug	CAAGGAACACCAUAUGAAG	-2.1	-2.1	-0.7	1	1	2	-34.6	73.9	3	5	II	42.1	60.2	99.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0739	NM_003344	Huesken	UBE2H	7328	GGUGUUCCUUGUGGUCCAUaa	AUGGACCACAAGGAACACC	-1.1	-3.3	-0.8	-1	-1	2	-40.2	47.4	-1	2	III	52.6	23.5	34	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0740	NM_003344	Huesken	UBE2H	7328	UGUUCCUUGUGGUCCAUAAaa	UUAUGGACCACAAGGAACA	-0.9	-2.1	-0.8	2	-1	2	-36.9	83.6	-3.8	7	II	42.1	50	91.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0741	NM_003344	Huesken	UBE2H	7328	GUGGUCCAUAAAACUUCACua	GUGAAGUUUUAUGGACCAC	-2.2	-2.2	0.5	0	1	2	-34.6	64.2	7.1	3	II	42.1	48.4	63.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0742	NM_003344	Huesken	UBE2H	7328	UGGUCCAUAAAACUUCACUac	AGUGAAGUUUUAUGGACCA	-2.1	-2.1	0.5	0	1	2	-34.5	87.2	3.1	6	II	36.8	64.9	95.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0743	NM_003344	Huesken	UBE2H	7328	UCCAUAAAACUUCACUACAaa	UGUAGUGAAGUUUUAUGGA	-2.1	-2.4	1.1	0	-1	2	-32.5	75.4	1.7	6	II	31.6	58.8	75.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0744	NM_003344	Huesken	UBE2H	7328	AAAACUUCACUACAAAUUCau	GAAUUUGUAGUGAAGUUUU	-2.4	-0.9	0	2	3	1	-28.5	87.7	9.8	7	Ia	26.3	74	96.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0745	NM_003344	Huesken	UBE2H	7328	CCUCCCAGGAUCGUAACCUca	AGGUUACGAUCCUGGGAGG	-2.1	-3.3	-1.2	1	0	3	-42.2	45.2	-6	1	III	57.9	43.2	84.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0746	NM_003344	Huesken	UBE2H	7328	CUCCCAGGAUCGUAACCUCau	GAGGUUACGAUCCUGGGAG	-2.4	-2.1	-1.6	0	0	3	-41.3	44	3.1	0	II	57.9	42.8	57.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0747	NM_003344	Huesken	UBE2H	7328	CCCAGGAUCGUAACCUCAUgu	AUGAGGUUACGAUCCUGGG	-1.1	-3.3	-1.6	0	-3	3	-40	42.6	-8.6	1	III	52.6	37.6	77.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0748	NM_003344	Huesken	UBE2H	7328	AACCUCAUGUUUACUCUCGau	CGAGAGUAAACAUGAGGUU	-2.4	-0.9	1.3	3	3	2	-35.1	75.7	0.4	6	Ib	42.1	66.1	98	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0749	NM_003344	Huesken	UBE2H	7328	CCUCAUGUUUACUCUCGAUga	AUCGAGAGUAAACAUGAGG	-1.1	-3.3	2.3	0	-2	2	-35.5	62.4	-3.3	3	III	42.1	34.8	71.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0750	NM_003344	Huesken	UBE2H	7328	CUCAUGUUUACUCUCGAUGag	CAUCGAGAGUAAACAUGAG	-2.1	-2.1	2.2	0	0	2	-34.3	82.1	3.4	5	II	42.1	60.1	70.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0751	NM_003344	Huesken	UBE2H	7328	UCAUGUUUACUCUCGAUGAgc	UCAUCGAGAGUAAACAUGA	-2.4	-2.4	1.4	1	0	2	-34.6	81.4	-6.1	7	II	36.8	62.1	71.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0752	NM_003345	Huesken	UBE2I	7329	UUUGGCAGUAAAUCGUGUAgg	UACACGAUUUACUGCCAAA	-1.3	-0.9	2.4	0	1	3	-33.8	75.7	-1.5	7	II	36.8	62.5	92.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0753	NM_003345	Huesken	UBE2I	7329	GGCAGUAAAUCGUGUAGGCcu	GCCUACACGAUUUACUGCC	-3.4	-3.3	1.9	-1	1	3	-38.7	51.3	5.1	2	II	52.6	38.9	67.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0754	NM_003345	Huesken	UBE2I	7329	UAAAUCGUGUAGGCCUCUGcu	CAGAGGCCUACACGAUUUA	-2.1	-1.3	1.5	1	4	4	-37.6	70.7	2.3	8	Ia	47.4	69.2	83.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0755	NM_003345	Huesken	UBE2I	7329	GUAGGCCUCUGCUUGAGCUgg	AGCUCAAGCAGAGGCCUAC	-2.1	-2.2	-1.9	1	2	4	-42.8	54.8	-4	0	III	57.9	42.1	89.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0756	NM_003345	Huesken	UBE2I	7329	CUGCUUGAGCUGGGUCUUGga	CAAGACCCAGCUCAAGCAG	-2.1	-2.1	-1.4	0	-1	3	-41.1	41.3	-2	2	II	57.9	44.9	75.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0757	NM_003345	Huesken	UBE2I	7329	UGGGUCUUGGAUAUUUGGUuc	ACCAAAUAUCCAAGACCCA	-2.2	-2.1	3	2	2	3	-37	79	1.1	5	II	42.1	60.6	83.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0758	NM_003345	Huesken	UBE2I	7329	GGGUCUUGGAUAUUUGGUUca	AACCAAAUAUCCAAGACCC	-0.9	-3.3	3	2	-1	3	-35.8	70	-1.6	3	III	42.1	40.8	48.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0759	NM_003345	Huesken	UBE2I	7329	UAUUUGGUUCAUUUAGAAGuu	CUUCUAAAUGAACCAAAUA	-2.1	-1.3	0.2	0	4	2	-28.9	101.7	7.7	8	Ia	26.3	71.5	88.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0760	NM_003345	Huesken	UBE2I	7329	CAUUUAGAAGUUCCUGUAUuc	AUACAGGAACUUCUAAAUG	-1.1	-2.1	0.9	-1	-2	2	-31.4	77.5	-0.6	6	II	31.6	42.9	52.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0761	NM_003345	Huesken	UBE2I	7329	AGUUCCUGUAUUCCUAAUAgg	UAUUAGGAAUACAGGAACU	-1.3	-2.1	2.4	2	-1	2	-33	85.7	1.7	6	II	31.6	51.8	71.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0762	NM_003345	Huesken	UBE2I	7329	UGUAUUCCUAAUAGGAUCUgu	AGAUCCUAUUAGGAAUACA	-2.1	-2.1	-2.4	-1	1	2	-33.4	89.6	-1.2	7	II	31.6	66.5	90.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0763	NM_003345	Huesken	UBE2I	7329	GUAUUCCUAAUAGGAUCUGuu	CAGAUCCUAUUAGGAAUAC	-2.1	-2.2	-2.4	-1	2	2	-33.4	63.9	0	4	II	36.8	48.2	74.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0764	NM_003345	Huesken	UBE2I	7329	UAGGAUCUGUUUGAUUGUGau	CACAAUCAAACAGAUCCUA	-2.1	-1.3	0.7	1	3	2	-33.7	71.4	5.7	5	Ib	36.8	64.8	87.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0765	NM_003345	Huesken	UBE2I	7329	CUGUUUGAUUGUGAUGGCUgg	AGCCAUCACAAUCAAACAG	-2.1	-2.1	4.4	-1	-1	3	-35.5	66.6	-3	4	II	42.1	47.9	60.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0766	NM_003345	Huesken	UBE2I	7329	GUUUGAUUGUGAUGGCUGGcc	CCAGCCAUCACAAUCAAAC	-3.3	-2.2	4.4	1	2	3	-36.7	64.4	0	5	II	47.4	46.2	56	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0767	NM_003345	Huesken	UBE2I	7329	UGGCCUCCAGUCCUUGUCCuc	GGACAAGGACUGGAGGCCA	-3.3	-2.1	-0.9	1	3	4	-45	61.5	9.8	3	II	63.2	63	85.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0768	NM_003345	Huesken	UBE2I	7329	UCCAGUCCUUGUCCUCCUCua	GAGGAGGACAAGGACUGGA	-2.4	-2.4	0.5	0	2	2	-43.1	62.5	12.8	3	II	57.9	56	70.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0769	NM_003345	Huesken	UBE2I	7329	UCUAAGAUGGACAGGCACAcu	UGUGCCUGUCCAUCUUAGA	-2.1	-2.4	0.3	0	0	3	-39.6	55.5	-3.7	7	II	47.4	56.8	76	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0770	NM_003345	Huesken	UBE2I	7329	AAGAUGGACAGGCACACUGuc	CAGUGUGCCUGUCCAUCUU	-2.1	-0.9	-0.2	2	3	3	-40.2	60.5	2.4	7	Ib	52.6	66	80.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0771	NM_003345	Huesken	UBE2I	7329	AGAUGGACAGGCACACUGUcc	ACAGUGUGCCUGUCCAUCU	-2.2	-2.1	-1.4	1	1	3	-41.5	50.7	-8.9	5	II	52.6	51.3	59.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0772	NM_003345	Huesken	UBE2I	7329	GGACAGGCACACUGUCCCCga	GGGGACAGUGUGCCUGUCC	-3.3	-3.3	-7.5	0	2	4	-46.1	33.9	10.1	1	II	68.4	41.5	47.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0773	NM_003345	Huesken	UBE2I	7329	GGCACACUGUCCCCGAAGGgu	CCUUCGGGGACAGUGUGCC	-3.3	-3.3	-3.8	1	0	5	-45.1	40.1	-2.3	-1	II	68.4	31.2	51.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0774	NM_003345	Huesken	UBE2I	7329	UCCCCGAAGGGUACACAUUcg	AAUGUGUACCCUUCGGGGA	-0.9	-2.4	-7.4	2	-1	5	-40.8	51.3	-3.5	3	II	52.6	50.6	59.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0775	NM_003345	Huesken	UBE2I	7329	CCCGAAGGGUACACAUUCGgg	CGAAUGUGUACCCUUCGGG	-2.4	-3.3	-2.8	0	-1	4	-39.9	38.5	-1.9	1	II	57.9	40.3	45.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0776	NM_003345	Huesken	UBE2I	7329	GGGUACACAUUCGGGUGAAau	UUCACCCGAAUGUGUACCC	-0.9	-3.3	-0.4	-1	-3	4	-39.7	37.5	-3.4	3	III	52.6	19	49.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0777	NM_003345	Huesken	UBE2I	7329	UACACAUUCGGGUGAAAUAau	UAUUUCACCCGAAUGUGUA	-1.3	-1.3	-0.4	1	-1	4	-34.2	76.5	-3.8	6	II	36.8	55.2	78.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0778	NM_003345	Huesken	UBE2I	7329	AUUCGGGUGAAAUAAUGGUgg	ACCAUUAUUUCACCCGAAU	-2.2	-1.1	0.5	3	3	4	-33.9	66.5	-1.7	6	II	36.8	58.8	88.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0779	NM_003345	Huesken	UBE2I	7329	GAAAUAAUGGUGGUUCGAAuu	UUCGAACCACCAUUAUUUC	-0.9	-2.4	2.7	1	0	2	-32.8	60.2	-1.4	5	II	36.8	38.9	59.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0780	NM_003345	Huesken	UBE2I	7329	AAAUAAUGGUGGUUCGAAUuu	AUUCGAACCACCAUUAUUU	-1.1	-0.9	3.8	3	2	2	-31.5	78.8	1.1	7	II	31.6	52.8	81	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0781	NM_003345	Huesken	UBE2I	7329	AAUGGUGGUUCGAAUUUACau	GUAAAUUCGAACCACCAUU	-2.2	-0.9	2.1	1	3	2	-32.6	69.1	2.5	6	Ib	36.8	60	89.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0782	NM_003345	Huesken	UBE2I	7329	UGGUGGUUCGAAUUUACAUuu	AUGUAAAUUCGAACCACCA	-1.1	-2.1	1.9	2	1	2	-33.8	81.3	0.7	5	II	36.8	59.4	75	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0783	NM_003345	Huesken	UBE2I	7329	GUGGUUCGAAUUUACAUUUug	AAAUGUAAAUUCGAACCAC	-0.9	-2.2	1.9	0	-2	2	-30.2	77.4	-1	4	III	31.6	50.1	53.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0784	NM_003345	Huesken	UBE2I	7329	UUCGAAUUUACAUUUUGGUgg	ACCAAAAUGUAAAUUCGAA	-2.2	-0.9	1.9	1	3	2	-28.9	90.7	-1.3	7	II	26.3	69.1	86.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0785	NM_003345	Huesken	UBE2I	7329	UGGCGAAGAUGGAUAAUCAuc	UGAUUAUCCAUCUUCGCCA	-2.1	-2.1	0.8	1	0	4	-36.8	58.6	-4	5	II	42.1	52.5	59	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0786	NM_003345	Huesken	UBE2I	7329	CGAAGAUGGAUAAUCAUCUuu	AGAUGAUUAUCCAUCUUCG	-2.1	-2.4	-3.4	1	-1	2	-33.6	63.8	-6.2	4	II	36.8	54.6	52.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0787	NM_003345	Huesken	UBE2I	7329	AUGGAUAAUCAUCUUUGAAaa	UUCAAAGAUGAUUAUCCAU	-0.9	-1.1	0.5	1	1	2	-30.6	82.1	9	5	II	26.3	52.2	65.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0788	NM_003345	Huesken	UBE2I	7329	CAUCUUUGAAAAGCAUCCGua	CGGAUGCUUUUCAAAGAUG	-2.4	-2.1	-0.3	1	2	3	-33.5	71.3	0.3	5	II	42.1	61.1	87.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0789	NM_003345	Huesken	UBE2I	7329	AAAAGCAUCCGUAGUUUAAac	UUAAACUACGGAUGCUUUU	-0.9	-0.9	2.9	1	0	3	-31.3	75.6	-3.3	7	II	31.6	49.2	50.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0790	NM_003345	Huesken	UBE2I	7329	AGCAUCCGUAGUUUAAACAag	UGUUUAAACUACGGAUGCU	-2.1	-2.1	1.2	1	0	3	-33.8	80	1.6	5	II	36.8	49.8	68.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0791	NM_003345	Huesken	UBE2I	7329	AUCCGUAGUUUAAACAAGCcu	GCUUGUUUAAACUACGGAU	-3.4	-1.1	0.6	2	3	3	-32.6	64.1	4.8	4	Ib	36.8	69.3	92.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0792	NM_003345	Huesken	UBE2I	7329	CGUAGUUUAAACAAGCCUCcu	GAGGCUUGUUUAAACUACG	-2.4	-2.4	1.2	0	0	3	-33.6	72.3	2.8	4	II	42.1	58.5	73.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0793	NM_003345	Huesken	UBE2I	7329	GUAGUUUAAACAAGCCUCCuu	GGAGGCUUGUUUAAACUAC	-3.3	-2.2	1.5	0	2	3	-34.5	66.3	2.5	6	II	42.1	63.9	82.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0794	NM_003345	Huesken	UBE2I	7329	UAAACAAGCCUCCUUCCCAcg	UGGGAAGGAGGCUUGUUUA	-2.1	-1.3	0.6	1	3	3	-39	67.8	-1.4	7	II	47.4	65.8	78	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0795	NM_003345	Huesken	UBE2I	7329	CAAGCCUCCUUCCCACGGAgu	UCCGUGGGAAGGAGGCUUG	-2.4	-2.1	-2.3	1	-1	3	-44	43.1	-5.7	1	III	63.2	39.7	52.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0796	NM_003345	Huesken	UBE2I	7329	UCCUUCCCACGGAGUCCCUuu	AGGGACUCCGUGGGAAGGA	-2.1	-2.4	-1.6	-1	1	3	-45.5	48.8	-6.6	3	II	63.2	45.8	72.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0797	NM_003345	Huesken	UBE2I	7329	UUCCCACGGAGUCCCUUUCuu	GAAAGGGACUCCGUGGGAA	-2.4	-0.9	-0.8	2	2	3	-41.9	66.5	12.8	5	II	57.9	61.1	89.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0798	NM_003345	Huesken	UBE2I	7329	UCCCACGGAGUCCCUUUCUuu	AGAAAGGGACUCCGUGGGA	-2.1	-2.4	-0.8	1	0	3	-43.1	51.6	-1.2	5	II	57.9	52.3	63.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0799	NM_003345	Huesken	UBE2I	7329	CCCACGGAGUCCCUUUCUUuc	AAGAAAGGGACUCCGUGGG	-0.9	-3.3	-0.8	0	-2	3	-41.6	44.9	-5.9	1	III	57.9	30.5	45.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0800	NM_003345	Huesken	UBE2I	7329	CACGGAGUCCCUUUCUUUCcu	GAAAGAAAGGGACUCCGUG	-2.4	-2.1	-0.8	0	0	3	-38.3	64.5	8.4	3	II	52.6	49.4	53.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0801	NM_003345	Huesken	UBE2I	7329	GAGUCCCUUUCUUUCCUGGaa	CCAGGAAAGAAAGGGACUC	-3.3	-2.4	0.4	0	2	3	-39.1	73.5	3	3	II	52.6	45.8	54.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0802	NM_003345	Huesken	UBE2I	7329	UCCCUUUCUUUCCUGGAAUgg	AUUCCAGGAAAGAAAGGGA	-1.1	-2.4	-0.9	2	0	3	-36.8	73.9	-1.6	5	II	42.1	53.8	83.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0803	NM_003345	Huesken	UBE2I	7329	UUCCUGGAAUGGCGCACUCcc	GAGUGCGCCAUUCCAGGAA	-2.4	-0.9	-1.6	0	3	5	-41.9	52.6	9.8	4	II	57.9	60.9	72.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0804	NM_003345	Huesken	UBE2I	7329	GCACUCCCAGUUCAUGAGGuu	CCUCAUGAACUGGGAGUGC	-3.3	-3.4	-1.8	1	1	3	-41.6	47.2	5.4	0	II	57.9	39.8	72.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0805	NM_003345	Huesken	UBE2I	7329	CACUCCCAGUUCAUGAGGUuc	ACCUCAUGAACUGGGAGUG	-2.2	-2.1	-1.8	0	0	3	-40.4	55	-8.6	1	III	52.6	38.2	58.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0806	NM_003345	Huesken	UBE2I	7329	ACUCCCAGUUCAUGAGGUUca	AACCUCAUGAACUGGGAGU	-0.9	-2.2	-1.8	1	1	3	-39.2	49.5	-6.3	3	II	47.4	38.9	34.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0807	NM_003345	Huesken	UBE2I	7329	CAGUUCAUGAGGUUCAUCGug	CGAUGAACCUCAUGAACUG	-2.4	-2.1	-0.3	0	1	2	-36.3	82.3	0.3	6	II	47.4	62.9	78.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0808	NM_003345	Huesken	UBE2I	7329	UUCAUGAGGUUCAUCGUGCca	GCACGAUGAACCUCAUGAA	-3.4	-0.9	-0.3	1	4	2	-37.6	72.4	4.8	8	Ia	47.4	75.9	84.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0809	NM_003345	Huesken	UBE2I	7329	AGGUUCAUCGUGCCAUCGGga	CCGAUGGCACGAUGAACCU	-3.3	-2.1	0.5	2	3	3	-41.2	59	2.4	4	Ib	57.9	57	81.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0810	NM_003345	Huesken	UBE2I	7329	GGUUCAUCGUGCCAUCGGGau	CCCGAUGGCACGAUGAACC	-3.3	-3.3	0.5	2	2	4	-42.4	50.3	0.1	2	II	63.2	37.7	55	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0811	NM_003345	Huesken	UBE2I	7329	GUUCAUCGUGCCAUCGGGAuu	UCCCGAUGGCACGAUGAAC	-2.4	-2.2	0.5	1	1	4	-41.5	43.5	-6	2	II	57.9	36.6	48.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0812	NM_003345	Huesken	UBE2I	7329	GACAGCCACGAAACCAAAUgg	AUUUGGUUUCGUGGCUGUC	-1.1	-2.4	-0.7	1	-2	3	-36.9	51.8	-4.3	2	III	47.4	35.6	51.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0813	NM_003345	Huesken	UBE2I	7329	GCCACGAAACCAAAUGGGUgg	ACCCAUUUGGUUUCGUGGC	-2.2	-3.4	-2	0	-1	3	-39	40.5	-8.9	1	III	52.6	33.1	56.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0814	NM_003345	Huesken	UBE2I	7329	CCACGAAACCAAAUGGGUGgu	CACCCAUUUGGUUUCGUGG	-2.1	-3.3	-1.6	1	1	3	-37.7	42.8	0	1	II	52.6	44.4	85.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0815	NM_003345	Huesken	UBE2I	7329	GAAACCAAAUGGGUGGUCUuu	AGACCACCCAUUUGGUUUC	-2.1	-2.4	-3.9	0	1	3	-37.7	63.9	-1.6	4	II	47.4	44.8	79.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0816	NM_014501	Huesken	UBE2S	27338	UUCUUAUCGCGCUCGCCAGca	CUGGCGAGCGCGAUAAGAA	-2.1	-0.9	-1.5	2	3	5	-40.5	69.8	-2.4	5	Ia	57.9	62.8	82.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0817	NM_014501	Huesken	UBE2S	27338	UUAUCGCGCUCGCCAGCAUgc	AUGCUGGCGAGCGCGAUAA	-1.1	-0.9	-2.3	1	2	5	-41.7	53.6	-3.9	4	II	57.9	50.4	71.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0818	NM_014501	Huesken	UBE2S	27338	GCCAGCAUGCUUCUUGGCCau	GGCCAAGAAGCAUGCUGGC	-3.3	-3.4	-2.8	1	2	4	-43.5	51.9	15.2	0	II	63.2	43.5	57.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0819	NM_014501	Huesken	UBE2S	27338	GCCAUGGGACCCUCAGCCCcu	GGGCUGAGGGUCCCAUGGC	-3.3	-3.4	-4.3	0	1	4	-48.5	34.8	2.7	1	II	73.7	40.3	45.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0820	NM_014501	Huesken	UBE2S	27338	CCAUGGGACCCUCAGCCCCuc	GGGGCUGAGGGUCCCAUGG	-3.3	-3.3	-8.5	0	1	5	-48.4	33.6	8.5	1	II	73.7	47.4	42.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0821	NM_014501	Huesken	UBE2S	27338	AUGGGACCCUCAGCCCCUCcc	GAGGGGCUGAGGGUCCCAU	-2.4	-1.1	-8.7	2	3	5	-47.5	32	5.1	1	II	68.4	51.7	45	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0822	NM_014501	Huesken	UBE2S	27338	GGCCCCAGGGUCGGUGGAGga	CUCCACCGACCCUGGGGCC	-2.1	-3.3	-3.7	1	0	6	-49.8	16.6	-2.4	-1	II	78.9	23	44	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0823	NM_014501	Huesken	UBE2S	27338	AGGGUCGGUGGAGGAAGCUuc	AGCUUCCUCCACCGACCCU	-2.1	-2.1	2.2	0	1	3	-45.3	29.9	-4	3	II	63.2	41.9	49.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0824	NM_014501	Huesken	UBE2S	27338	GUGGAGGAAGCUUCAGUGCca	GCACUGAAGCUUCCUCCAC	-3.4	-2.2	-0.7	0	1	2	-41.5	48.9	10.8	3	II	57.9	48.7	58.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0825	NM_014501	Huesken	UBE2S	27338	AAGCUUCAGUGCCACUGGCca	GCCAGUGGCACUGAAGCUU	-3.4	-0.9	-2.7	3	4	3	-42.2	54.5	7.5	2	Ib	57.9	56.2	57.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0826	NM_014501	Huesken	UBE2S	27338	GUGCCACUGGCCAGGGCCCga	GGGCCCUGGCCAGUGGCAC	-3.3	-2.2	-5.7	1	2	6	-50.3	24.1	0.1	-1	II	78.9	39.1	43	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0827	NM_014501	Huesken	UBE2S	27338	ACUGGCCAGGGCCCGACCGgc	CGGUCGGGCCCUGGCCAGU	-2.4	-2.2	-7.1	2	4	7	-49.9	31.2	0.1	1	II	78.9	48.4	41.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0828	NM_014501	Huesken	UBE2S	27338	GGCCAGGGCCCGACCGGCUuc	AGCCGGUCGGGCCCUGGCC	-2.1	-3.3	-6.3	2	-1	7	-52.3	12	-6.2	0	III	84.2	27.5	43	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0829	NM_014501	Huesken	UBE2S	27338	GCCCGACCGGCUUCGGCCCug	GGGCCGAAGCCGGUCGGGC	-3.3	-3.4	-7.3	2	1	6	-50.5	23.4	8.1	-2	II	84.2	31.7	36.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0830	NM_014501	Huesken	UBE2S	27338	CCCGACCGGCUUCGGCCCUgc	AGGGCCGAAGCCGGUCGGG	-2.1	-3.3	-7.3	-1	-2	6	-49.2	17.5	2.2	-1	III	78.9	24.8	49.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0831	NM_014501	Huesken	UBE2S	27338	UUCGGCCCUGCCGCUGGGCcc	GCCCAGCGGCAGGGCCGAA	-3.4	-0.9	-6.4	0	4	6	-49.9	35.5	2.7	2	II	78.9	57	42	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0832	NM_014501	Huesken	UBE2S	27338	CGUGGAUCUCUGUGAGCAGac	CUGCUCACAGAGAUCCACG	-2.1	-2.4	-0.8	-1	0	2	-41	39.9	-2	1	II	57.9	36.2	54.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0833	NM_014501	Huesken	UBE2S	27338	AUCUCUGUGAGCAGACGGGcc	CCCGUCUGCUCACAGAGAU	-3.3	-1.1	-2.6	1	4	4	-42.2	56.5	-4.9	4	Ib	57.9	57.9	70.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0834	NM_014501	Huesken	UBE2S	27338	GUGAGCAGACGGGCCCGAGcu	CUCGGGCCCGUCUGCUCAC	-2.1	-2.2	-4.4	0	1	8	-46.9	31.2	0	1	II	73.7	36.9	55	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0835	NM_014501	Huesken	UBE2S	27338	ACGGGCCCGAGCUGCAUACuc	GUAUGCAGCUCGGGCCCGU	-2.2	-2.2	-0.6	1	1	8	-45.8	40.1	5.1	3	II	68.4	43.7	47	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0836	NM_014501	Huesken	UBE2S	27338	CGGGCCCGAGCUGCAUACUcc	AGUAUGCAGCUCGGGCCCG	-2.1	-2.4	-0.6	-1	-3	8	-45.7	32.8	-5.3	-1	III	68.4	28.8	49.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0837	NM_014501	Huesken	UBE2S	27338	GGGCCCGAGCUGCAUACUCcu	GAGUAUGCAGCUCGGGCCC	-2.4	-3.3	-3.6	2	0	7	-45.7	32.9	9.8	-1	II	68.4	36.8	53.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0838	NM_014501	Huesken	UBE2S	27338	CCCGAGCUGCAUACUCCUCgu	GAGGAGUAUGCAGCUCGGG	-2.4	-3.3	-3.6	0	-1	4	-43.5	40	10.8	-1	II	63.2	38.3	56.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0839	NM_014501	Huesken	UBE2S	27338	CUGCAUACUCCUCGUAGUUcu	AACUACGAGGAGUAUGCAG	-0.9	-2.1	-0.3	-1	-2	2	-37.7	54.5	-0.2	3	II	47.4	41.4	40	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0840	NM_014501	Huesken	UBE2S	27338	AUACUCCUCGUAGUUCUCCaa	GGAGAACUACGAGGAGUAU	-3.3	-1.1	-0.3	3	5	2	-38.2	68.7	9.7	6	Ib	47.4	69	56.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0841	NM_014501	Huesken	UBE2S	27338	UCGUAGUUCUCCAAGAGCAgg	UGCUCUUGGAGAACUACGA	-2.1	-2.4	-0.1	2	0	2	-38.8	60.3	-8.3	6	II	47.4	58.1	77.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0842	NM_014501	Huesken	UBE2S	27338	GUAGUUCUCCAAGAGCAGGcg	CCUGCUCUUGGAGAACUAC	-3.3	-2.2	-0.5	1	2	2	-39.4	53.4	2.3	3	II	52.6	51.5	70.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0843	NM_014501	Huesken	UBE2S	27338	UAGUUCUCCAAGAGCAGGCgg	GCCUGCUCUUGGAGAACUA	-3.4	-1.3	-0.7	1	4	3	-40.6	74	7.1	8	Ia	52.6	74.5	76	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0844	NM_014501	Huesken	UBE2S	27338	UCUCCAAGAGCAGGCGGCCcg	GGCCGCCUGCUCUUGGAGA	-3.3	-2.4	-0.7	0	4	7	-46.5	34.9	5.1	4	Ib	68.4	56.1	69.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0845	NM_014501	Huesken	UBE2S	27338	AGGCGGCCCGCCUCCUCGUug	ACGAGGAGGCGGGCCGCCU	-2.2	-2.1	-5.4	3	1	11	-50.5	27.4	-8.7	0	II	78.9	39.7	44.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0846	NM_014501	Huesken	UBE2S	27338	CCGCCUCCUCGUUGAGUGCag	GCACUCAACGAGGAGGCGG	-3.4	-3.3	-2.3	-1	-1	5	-44.5	43.7	8.4	-1	II	68.4	35.1	48.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0847	NM_014501	Huesken	UBE2S	27338	CCUCGUUGAGUGCAGACUCgg	GAGUCUGCACUCAACGAGG	-2.4	-3.3	-2.9	1	0	2	-40.9	46.9	7.8	0	II	57.9	48.2	71.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0848	NM_014501	Huesken	UBE2S	27338	CUCGUUGAGUGCAGACUCGgg	CGAGUCUGCACUCAACGAG	-2.4	-2.1	-3.2	-1	0	2	-40	45.7	-6.9	2	II	57.9	48.9	70.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0849	NM_014501	Huesken	UBE2S	27338	GAGUGCAGACUCGGGGUUAgg	UAACCCCGAGUCUGCACUC	-1.3	-2.4	-1.5	-1	-3	5	-42.2	36.5	-6.1	2	III	57.9	28.4	39.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0850	NM_014501	Huesken	UBE2S	27338	GACUCGGGGUUAGGGUGGAuc	UCCACCCUAACCCCGAGUC	-2.4	-2.4	2.3	0	-1	5	-44.5	27.2	-3.8	2	III	63.2	23.5	39.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0851	NM_014501	Huesken	UBE2S	27338	GGGUGGAUCAGCAGGCACUug	AGUGCCUGCUGAUCCACCC	-2.1	-3.3	0.1	0	-2	3	-45	39.4	-8.9	2	III	63.2	37.7	55.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0852	NM_014501	Huesken	UBE2S	27338	GUGGAUCAGCAGGCACUUGau	CAAGUGCCUGCUGAUCCAC	-2.1	-2.2	-0.2	1	0	3	-41.4	41.9	5	2	II	57.9	41.3	63.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0853	NM_014501	Huesken	UBE2S	27338	GAUGGUCAGCAGUACGUGUcg	ACACGUACUGCUGACCAUC	-2.2	-2.4	1.7	1	0	2	-39.9	53.2	-1.7	3	III	52.6	39.9	59.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0854	NM_014501	Huesken	UBE2S	27338	AUGGUCAGCAGUACGUGUCgg	GACACGUACUGCUGACCAU	-2.4	-1.1	1.7	2	3	2	-39.9	67.5	7.8	6	Ib	52.6	65.4	85.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0855	NM_014501	Huesken	UBE2S	27338	GGAUGCCCAGCUCAGCCGUcc	ACGGCUGAGCUGGGCAUCC	-2.2	-3.3	-3	0	0	4	-46.5	36.4	-0.9	0	III	68.4	30.1	52.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0856	NM_014501	Huesken	UBE2S	27338	CCUCUUGAGCACGUUGACGca	CGUCAACGUGCUCAAGAGG	-2.4	-3.3	0.7	1	1	2	-39.7	51.7	0.3	2	II	57.9	47	63.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0857	NM_014501	Huesken	UBE2S	27338	UCUUGAGCACGUUGACGCAga	UGCGUCAACGUGCUCAAGA	-2.1	-2.4	-1.6	-1	1	3	-39.8	54.3	-0.7	5	II	52.6	45.5	74.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0858	NM_014501	Huesken	UBE2S	27338	CUUGAGCACGUUGACGCAGau	CUGCGUCAACGUGCUCAAG	-2.1	-2.1	-1.6	0	1	3	-39.5	41.3	3.7	1	II	57.9	46.6	67.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0859	NM_014501	Huesken	UBE2S	27338	GUUGACGCAGAUCUCGCCAuu	UGGCGAGAUCUGCGUCAAC	-2.1	-2.2	-4.2	0	1	4	-41.4	54.1	9	2	III	57.9	42	85.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0860	NM_014501	Huesken	UBE2S	27338	UCUCGCCAUUGGCGCCCACgu	GUGGGCGCCAAUGGCGAGA	-2.2	-2.4	-4.7	0	3	7	-45.6	38.3	9.8	2	II	68.4	45.1	54.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0861	NM_014501	Huesken	UBE2S	27338	CAUUGGCGCCCACGUUCGGgu	CCGAACGUGGGCGCCAAUG	-3.3	-2.1	-0.7	-1	2	7	-43.2	36.3	-2	2	II	68.4	39.2	44.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0862	NM_014501	Huesken	UBE2S	27338	CCACGUUCGGGUGGAAGAUcu	AUCUUCCACCCGAACGUGG	-1.1	-3.3	-3.2	1	-2	4	-41	35.4	-3	0	III	57.9	29.3	31.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0863	NM_014501	Huesken	UBE2S	27338	CGUUCGGGUGGAAGAUCUUgg	AAGAUCUUCCACCCGAACG	-0.9	-2.4	1.2	-2	-2	4	-38.8	41.2	-8.6	3	III	52.6	32.1	60.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0864	NM_014501	Huesken	UBE2S	27338	UCUUGGUCAGGAAGUAGCCcu	GGCUACUUCCUGACCAAGA	-3.3	-2.4	0.4	0	4	3	-40.6	59.5	4.8	6	Ib	52.6	65.4	91.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0865	NM_014501	Huesken	UBE2S	27338	AGGAAGUAGCCCUUGGGUGgg	CACCCAAGGGCUACUUCCU	-2.1	-2.1	-3.5	3	3	4	-42.4	54.3	0.3	3	Ib	57.9	53.9	81.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0866	NM_014501	Huesken	UBE2S	27338	GGAAGUAGCCCUUGGGUGGgg	CCACCCAAGGGCUACUUCC	-3.3	-3.3	-1.8	0	1	4	-43.6	48.1	0.7	2	II	63.2	42.5	54.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0867	NM_014501	Huesken	UBE2S	27338	AGUAGCCCUUGGGUGGGGAgg	UCCCCACCCAAGGGCUACU	-2.4	-2.1	-3.4	1	2	4	-46	43.1	-1.4	2	II	63.2	34.5	48.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0868	NM_014501	Huesken	UBE2S	27338	AGCCCUUGGGUGGGGAGGCag	GCCUCCCCACCCAAGGGCU	-3.4	-2.1	-3.7	3	3	4	-49.2	32.5	7.4	1	II	73.7	47.4	46.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0869	NM_014501	Huesken	UBE2S	27338	GGGGAGGCAGGGAAGUCCUuc	AGGACUUCCCUGCCUCCCC	-2.1	-3.3	-1.5	0	-1	4	-47.3	26.1	-3.9	0	III	68.4	34.5	55.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0870	NM_014501	Huesken	UBE2S	27338	CAGGGAAGUCCUUCCCCAGca	CUGGGGAAGGACUUCCCUG	-2.1	-2.1	-8	-1	0	4	-43.6	42.3	-1.3	0	II	63.2	44	40.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0871	NM_014501	Huesken	UBE2S	27338	AGGGAAGUCCUUCCCCAGCag	GCUGGGGAAGGACUUCCCU	-3.4	-2.1	-7.3	2	2	4	-44.9	45.9	15.5	2	II	63.2	54.8	53.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0872	NM_014501	Huesken	UBE2S	27338	UUCCCCAGCAGGAGUUUCAug	UGAAACUCCUGCUGGGGAA	-2.1	-0.9	-4.9	1	0	4	-41.3	62.8	-6.3	6	II	52.6	57	54	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0873	NM_014501	Huesken	UBE2S	27338	GAGUUUCAUGCGGAACAGAcc	UCUGUUCCGCAUGAAACUC	-2.4	-2.4	1	0	-2	4	-37.4	48.8	-6.1	3	II	47.4	35.5	42.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0874	NM_014501	Huesken	UBE2S	27338	CAUGCGGAACAGACCUCCAgc	UGGAGGUCUGUUCCGCAUG	-2.1	-2.1	-2.7	-1	-1	4	-41.9	44.9	-6	1	III	57.9	42	58.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0875	NM_014501	Huesken	UBE2S	27338	GACCUCCAGCAUAUGGGGUcc	ACCCCAUAUGCUGGAGGUC	-2.2	-2.4	-1.5	2	1	4	-43.1	51	1.1	0	III	57.9	33.5	49.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0876	NM_014501	Huesken	UBE2S	27338	CCAGCAUAUGGGGUCCCCUca	AGGGGACCCCAUAUGCUGG	-2.1	-3.3	-2.8	0	0	4	-45.1	39	-3.6	0	III	63.2	41.3	36.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0877	NM_014501	Huesken	UBE2S	27338	CAUAUGGGGUCCCCUCAGGgc	CCUGAGGGGACCCCAUAUG	-3.3	-2.1	-2.8	0	1	4	-44.1	47.6	-4.3	3	II	63.2	48.1	25.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0878	NM_014501	Huesken	UBE2S	27338	AUAUGGGGUCCCCUCAGGGcc	CCCUGAGGGGACCCCAUAU	-3.3	-1.1	-3.9	1	6	4	-45.3	49.3	0	4	Ib	63.2	61.7	46.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0879	NM_014501	Huesken	UBE2S	27338	GGGUCCCCUCAGGGCCCUCga	GAGGGCCCUGAGGGGACCC	-2.4	-3.3	-4.1	-1	0	6	-50.9	27.3	9.7	-1	II	78.9	28.4	44.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0880	NM_014501	Huesken	UBE2S	27338	UCAGGGCCCUCGAUGGUGAcc	UCACCAUCGAGGGCCCUGA	-2.4	-2.4	-1	1	1	6	-45.7	40.3	-6.4	3	II	63.2	46.8	35.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0881	NM_014501	Huesken	UBE2S	27338	CAGGGCCCUCGAUGGUGACcu	GUCACCAUCGAGGGCCCUG	-2.2	-2.1	-1	-2	0	6	-45.5	35.1	5.4	-1	II	68.4	33.7	45.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0882	NM_014501	Huesken	UBE2S	27338	UGAGGUCCUCCUCGUUGGGaa	CCCAACGAGGAGGACCUCA	-3.3	-2.1	-5.3	-1	4	3	-44.3	48.2	3.1	2	II	63.2	53.7	54	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0883	NM_014501	Huesken	UBE2S	27338	GAGGUCCUCCUCGUUGGGAaa	UCCCAACGAGGAGGACCUC	-2.4	-2.4	-3.6	1	0	3	-44.6	41.2	-3.8	0	III	63.2	31.5	35.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0884	NM_014501	Huesken	UBE2S	27338	AGGUCCUCCUCGUUGGGAAag	UUCCCAACGAGGAGGACCU	-0.9	-2.1	-2.2	3	0	3	-43.1	60.5	-3.8	2	II	57.9	36.3	41.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0885	NM_014501	Huesken	UBE2S	27338	GUCCUCCUCGUUGGGAAAGac	CUUUCCCAACGAGGAGGAC	-2.1	-2.2	-2.8	1	0	3	-40.7	47.5	3	0	II	57.9	36.2	33.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0886	NM_014501	Huesken	UBE2S	27338	CCUCGUUGGGAAAGACCUUga	AAGGUCUUUCCCAACGAGG	-0.9	-3.3	-0.8	1	-2	3	-39.2	37.4	-6.3	2	III	52.6	35.4	53	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0887	NM_014501	Huesken	UBE2S	27338	UGGGAAAGACCUUGAUGCCau	GGCAUCAAGGUCUUUCCCA	-3.3	-2.1	-1.2	-1	3	3	-40.3	54.3	10.8	4	Ib	52.6	59.1	57.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0888	NM_014501	Huesken	UBE2S	27338	GGAAAGACCUUGAUGCCAUcg	AUGGCAUCAAGGUCUUUCC	-1.1	-3.3	-1.2	1	0	3	-38.1	49.6	-1.5	3	II	47.4	38.6	56.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0889	NM_014501	Huesken	UBE2S	27338	GACCUUGAUGCCAUCGGGUgg	ACCCGAUGGCAUCAAGGUC	-2.2	-2.4	0	1	1	4	-42.1	47.3	-8.9	2	III	57.9	41.6	66.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0890	NM_014501	Huesken	UBE2S	27338	CCUUGAUGCCAUCGGGUGGgu	CCACCCGAUGGCAUCAAGG	-3.3	-3.3	-0.3	0	0	4	-42.9	50	5.7	3	II	63.2	39.1	47.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0891	NM_014501	Huesken	UBE2S	27338	UGAUGCCAUCGGGUGGGUCug	GACCCACCCGAUGGCAUCA	-2.4	-2.1	-0.6	0	4	4	-44.5	51.2	9.7	4	II	63.2	54.3	61.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0892	NM_014501	Huesken	UBE2S	27338	GCCAUCGGGUGGGUCUGCGgu	CGCAGACCCACCCGAUGGC	-2.4	-3.4	-0.6	1	2	4	-46.8	35.7	2.3	2	II	73.7	35	51.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0893	NM_014501	Huesken	UBE2S	27338	GGGUGGGUCUGCGGUCAGUgu	ACUGACCGCAGACCCACCC	-2.2	-3.3	-1.5	0	-2	4	-46.4	27.5	-6.3	0	III	68.4	24.4	43.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0894	NM_014501	Huesken	UBE2S	27338	UGCGGUCAGUGUCGUCACCuc	GGUGACGACACUGACCGCA	-3.3	-2.1	-2.1	0	2	4	-43.5	49.1	12.8	1	II	63.2	52.9	88.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0895	NM_004223	Huesken	UBE2L6	9246	UCCGAAUCGGAGGGUGAACuc	GUUCACCCUCCGAUUCGGA	-2.2	-2.4	-2	1	1	3	-41.5	44.4	12.4	3	Ib	57.9	49.5	61.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0896	NM_004223	Huesken	UBE2L6	9246	CCGAAUCGGAGGGUGAACUcu	AGUUCACCCUCCGAUUCGG	-2.1	-3.3	-0.1	0	-2	3	-41.2	46.3	-0.9	2	III	57.9	39.2	55.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0897	NM_004223	Huesken	UBE2L6	9246	CGGAGGGUGAACUCUUCGGca	CCGAAGAGUUCACCCUCCG	-3.3	-2.4	-1.1	0	0	3	-42.2	49.2	-2.1	1	II	63.2	46.8	62.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0898	NM_004223	Huesken	UBE2L6	9246	GAGGGUGAACUCUUCGGCAuu	UGCCGAAGAGUUCACCCUC	-2.1	-2.4	-0.3	0	0	4	-42	44.8	-3.8	1	III	57.9	40.2	59.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0899	NM_004223	Huesken	UBE2L6	9246	GGGUGAACUCUUCGGCAUUcu	AAUGCCGAAGAGUUCACCC	-0.9	-3.3	0.9	-1	-3	4	-39.5	43.4	1.5	2	III	52.6	29.2	51.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0900	NM_004223	Huesken	UBE2L6	9246	CUUCGGCAUUCUUUCUGAAca	UUCAGAAAGAAUGCCGAAG	-0.9	-2.1	0.7	0	-1	4	-34.8	60.9	-2.8	2	III	42.1	32.8	57.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0901	NM_004223	Huesken	UBE2L6	9246	AUUCUUUCUGAACAGCUCCgg	GGAGCUGUUCAGAAAGAAU	-3.3	-1.1	-0.3	3	4	2	-35.8	81.5	12.1	7	Ia	42.1	76	107.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0902	NM_004223	Huesken	UBE2L6	9246	AACAGCUCCGGAUUCUGUGuc	CACAGAAUCCGGAGCUGUU	-2.1	-0.9	-1.2	3	4	4	-39.5	70.4	2.3	4	Ib	52.6	61.2	97.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0903	NM_004223	Huesken	UBE2L6	9246	ACAGCUCCGGAUUCUGUGUca	ACACAGAAUCCGGAGCUGU	-2.2	-2.2	-1.1	2	1	4	-40.8	63	1.4	3	II	52.6	47.5	85.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0904	NM_004223	Huesken	UBE2L6	9246	AGCUCCGGAUUCUGUGUCAgc	UGACACAGAAUCCGGAGCU	-2.1	-2.1	2.6	0	-1	4	-41	55.2	-6.1	5	II	52.6	37.8	54.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0905	NM_004223	Huesken	UBE2L6	9246	GCUCCGGAUUCUGUGUCAGca	CUGACACAGAAUCCGGAGC	-2.1	-3.4	0.4	1	2	4	-41	45.8	3	1	II	57.9	32.6	79.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0906	NM_004223	Huesken	UBE2L6	9246	GUCAGCAGGUCAGCGAGGUcc	ACCUCGCUGACCUGCUGAC	-2.2	-2.2	-1.5	1	0	3	-44.3	34.5	-6.3	1	III	63.2	34.8	58.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0907	NM_004223	Huesken	UBE2L6	9246	GCAGGUCAGCGAGGUCCAUcc	AUGGACCUCGCUGACCUGC	-1.1	-3.4	-0.3	0	-1	3	-44.4	19.9	-4	0	III	63.2	19.9	43.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0908	NM_004223	Huesken	UBE2L6	9246	AGGUCAGCGAGGUCCAUCCgc	GGAUGGACCUCGCUGACCU	-3.3	-2.1	0	2	2	3	-44.6	59.9	7.4	5	II	63.2	58.7	80.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0909	NM_004223	Huesken	UBE2L6	9246	GGUCAGCGAGGUCCAUCCGca	CGGAUGGACCUCGCUGACC	-2.4	-3.3	-0.3	1	2	3	-44.9	37.6	8.1	1	II	68.4	36	71.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0910	NM_004223	Huesken	UBE2L6	9246	AGCGAGGUCCAUCCGCAGGgg	CCUGCGGAUGGACCUCGCU	-3.3	-2.1	-3.1	2	2	4	-45.8	45	10.4	2	II	68.4	50.4	81.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0911	NM_004223	Huesken	UBE2L6	9246	GCGAGGUCCAUCCGCAGGGgc	CCCUGCGGAUGGACCUCGC	-3.3	-3.4	-1.9	1	1	4	-47	31.1	0.1	1	II	73.7	42.4	59.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0912	NM_004223	Huesken	UBE2L6	9246	GGUCCAUCCGCAGGGGCUCcc	GAGCCCCUGCGGAUGGACC	-2.4	-3.3	-0.8	2	1	5	-47.9	27.8	5.1	1	II	73.7	33.9	55.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0913	NM_004223	Huesken	UBE2L6	9246	CCAUCCGCAGGGGCUCCCUga	AGGGAGCCCCUGCGGAUGG	-2.1	-3.3	-3.2	-1	-1	5	-48.7	33.8	-6	1	III	73.7	34.7	60.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0914	NM_004223	Huesken	UBE2L6	9246	GGGGCUCCCUGAUAUUCGGuc	CCGAAUAUCAGGGAGCCCC	-3.3	-3.3	-1.6	1	1	5	-43.5	46.6	0	-2	II	63.2	37	50	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0915	NM_004223	Huesken	UBE2L6	9246	UGAUAUUCGGUCUAUUCACca	GUGAAUAGACCGAAUAUCA	-2.2	-2.1	1.5	2	3	3	-33.7	82.1	7.8	7	Ia	36.8	70.7	89.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0916	NM_004223	Huesken	UBE2L6	9246	CGGUCUAUUCACCAGCACAuu	UGUGCUGGUGAAUAGACCG	-2.1	-2.4	-2.5	-1	-4	3	-39.7	58.4	-3.4	1	III	52.6	40.7	59	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0917	NM_004223	Huesken	UBE2L6	9246	CUAUUCACCAGCACAUUGAgg	UCAAUGUGCUGGUGAAUAG	-2.4	-2.1	1.2	0	-2	2	-35.9	71.5	-10.7	7	II	42.1	51	69.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0918	NM_004223	Huesken	UBE2L6	9246	GCACAUUGAGGGCCUCCAGga	CUGGAGGCCCUCAAUGUGC	-2.1	-3.4	-0.5	2	1	5	-43.7	42.1	5.1	2	II	63.2	37.4	48.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0919	NM_004223	Huesken	UBE2L6	9246	ACAUUGAGGGCCUCCAGGAcu	UCCUGGAGGCCCUCAAUGU	-2.4	-2.2	-0.5	2	1	5	-43.9	47.3	-8.5	5	II	57.9	47.7	74.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0920	NM_004223	Huesken	UBE2L6	9246	CCUCCAGGACUUGGCAAGUcu	ACUUGCCAAGUCCUGGAGG	-2.2	-3.3	-1.6	-1	-2	3	-42.3	34.4	-3	0	III	57.9	29.9	31.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0921	NM_004223	Huesken	UBE2L6	9246	UGGCAAGUCUUGGUGCAAGgc	CUUGCACCAAGACUUGCCA	-2.1	-2.1	-0.9	2	2	3	-39.7	51.9	5	5	II	52.6	54	89.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0922	NM_004223	Huesken	UBE2L6	9246	UCUUGGUGCAAGGCUUCCAgu	UGGAAGCCUUGCACCAAGA	-2.1	-2.4	-0.8	2	1	3	-41.1	62.2	-1.5	7	II	52.6	57	77.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0923	NM_004223	Huesken	UBE2L6	9246	UGGUGCAAGGCUUCCAGUUcu	AACUGGAAGCCUUGCACCA	-0.9	-2.1	0.4	1	0	3	-40.9	61.6	-3.3	5	II	52.6	54.6	77.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0924	NM_004223	Huesken	UBE2L6	9246	CUUCCAGUUCUCACUGCUGau	CAGCAGUGAGAACUGGAAG	-2.1	-2.1	-3.7	0	1	2	-39	53.8	-6.9	2	II	52.6	50.3	74.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0925	NM_004223	Huesken	UBE2L6	9246	UCCAGUUCUCACUGCUGAUga	AUCAGCAGUGAGAACUGGA	-1.1	-2.4	-3.7	0	0	2	-39.5	66.3	-4	4	II	47.4	49.8	82.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0926	NM_004223	Huesken	UBE2L6	9246	CCAGUUCUCACUGCUGAUGau	CAUCAGCAGUGAGAACUGG	-2.1	-3.3	-3.7	0	-1	2	-39.2	57.6	-1.3	2	II	52.6	46.3	65.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0927	NM_004223	Huesken	UBE2L6	9246	CAGUUCUCACUGCUGAUGAug	UCAUCAGCAGUGAGAACUG	-2.4	-2.1	-2.4	0	-2	2	-38.3	70.2	-1	4	II	47.4	44.8	61.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0928	NM_004223	Huesken	UBE2L6	9246	UCUCACUGCUGAUGAUGGGca	CCCAUCAUCAGCAGUGAGA	-3.3	-2.4	-1.1	2	4	3	-40.8	58	2.7	5	Ib	52.6	55.8	74	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0929	NM_004223	Huesken	UBE2L6	9246	GAUGAUGGGCAGGCAAAUCug	GAUUUGCCUGCCCAUCAUC	-2.4	-2.4	1.2	1	1	4	-39.5	48.6	12.4	4	II	52.6	46.9	71.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0930	NM_004223	Huesken	UBE2L6	9246	CUCGUCCACGUUGGGGUGGua	CCACCCCAACGUGGACGAG	-3.3	-2.1	-4.2	-1	0	4	-44.2	38.9	1	1	II	68.4	41.6	67.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0931	NM_004223	Huesken	UBE2L6	9246	UCGUCCACGUUGGGGUGGUag	ACCACCCCAACGUGGACGA	-2.2	-2.4	-4.7	0	1	4	-44.3	42.6	-4	3	II	63.2	44.4	64.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0932	NM_004223	Huesken	UBE2L6	9246	CGUCCACGUUGGGGUGGUAga	UACCACCCCAACGUGGACG	-1.3	-2.4	-4.7	-1	-3	4	-43.2	25.8	-5.8	-1	III	63.2	13.8	37.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0933	NM_004223	Huesken	UBE2L6	9246	UCCACGUUGGGGUGGUAGAuc	UCUACCACCCCAACGUGGA	-2.4	-2.4	-4.7	1	-1	4	-43.1	56	-3.8	3	II	57.9	45.8	53	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0934	NM_004223	Huesken	UBE2L6	9246	UGGGGUGGUAGAUCUUGGUug	ACCAAGAUCUACCACCCCA	-2.2	-2.1		0	1	4	-41.7	58.1	-4	5	II	52.6	55.9	90.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0935	NM_004223	Huesken	UBE2L6	9246	GGGGUGGUAGAUCUUGGUUgu	AACCAAGAUCUACCACCCC	-0.9	-3.3		0	-1	4	-40.5	54.2	6.1	2	III	52.6	33.3	43.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0936	NM_004223	Huesken	UBE2L6	9246	GUGGUAGAUCUUGGUUGUGaa	CACAACCAAGAUCUACCAC	-2.1	-2.2		-1	1	2	-37	47.6	3	3	II	47.4	41.2	45.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0937	NM_004223	Huesken	UBE2L6	9246	GUAGAUCUUGGUUGUGAAUuu	AUUCACAACCAAGAUCUAC	-1.1	-2.2	1.4	-1	-1	2	-33.8	66.6	1.8	3	II	36.8	37	51	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0938	NM_004223	Huesken	UBE2L6	9246	GAUCUUGGUUGUGAAUUUGau	CAAAUUCACAACCAAGAUC	-2.1	-2.4	1.4	1	1	2	-32.1	70.4	3	5	II	36.8	48.2	85.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0939	NM_004223	Huesken	UBE2L6	9246	UUGUGAAUUUGAUCAUGGGag	CCCAUGAUCAAAUUCACAA	-3.3	-0.9	-0.5	0	4	3	-33.4	78.4	0	6	Ia	36.8	68.6	95.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0940	NM_004223	Huesken	UBE2L6	9246	UUGAUCAUGGGAGGCUUGAac	UCAAGCCUCCCAUGAUCAA	-2.4	-0.9	1.5	0	0	3	-39.5	62.5	-3.8	8	II	47.4	58	86.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0941	NM_004223	Huesken	UBE2L6	9246	UGAUCAUGGGAGGCUUGAAcg	UUCAAGCCUCCCAUGAUCA	-0.9	-2.1	1.3	2	0	3	-39.5	60	-1.5	6	II	47.4	44.7	71.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0942	NM_004223	Huesken	UBE2L6	9246	GGAGGCUUGAACGGAUACUcc	AGUAUCCGUUCAAGCCUCC	-2.1	-3.3	1.4	1	-1	3	-39.9	44.8	-4	3	III	52.6	37.9	53.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0943	NM_004223	Huesken	UBE2L6	9246	AGGCUUGAACGGAUACUCCgg	GGAGUAUCCGUUCAAGCCU	-3.3	-2.1	-1.5	2	3	3	-39.9	60	7.1	4	II	52.6	59.9	84.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0944	NM_004223	Huesken	UBE2L6	9246	ACGGAUACUCCGGCGGGAAgc	UUCCCGCCGGAGUAUCCGU	-0.9	-2.2	-3.1	1	-1	8	-44.1	32.3	-3.4	2	II	63.2	34.7	43	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0945	NM_004223	Huesken	UBE2L6	9246	CGGAUACUCCGGCGGGAAGcu	CUUCCCGCCGGAGUAUCCG	-2.1	-2.4	-2.5	-1	-2	8	-44	37.1	0.4	0	II	68.4	28.2	38.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0946	NM_004223	Huesken	UBE2L6	9246	ACUCCGGCGGGAAGCUGAUgc	AUCAGCUUCCCGCCGGAGU	-1.1	-2.2	-2.5	2	1	8	-44.6	31.1	-6.6	2	II	63.2	34.7	52.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0947	NM_004223	Huesken	UBE2L6	9246	GGGAAGCUGAUGCGCAGGUug	ACCUGCGCAUCAGCUUCCC	-2.2	-3.3	-1.4	0	-1	4	-44.1	32.1	1.1	1	III	63.2	30.6	70.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0948	NM_004223	Huesken	UBE2L6	9246	GGAAGCUGAUGCGCAGGUUga	AACCUGCGCAUCAGCUUCC	-0.9	-3.3	-1.4	1	-1	4	-41.7	39.5	-6.3	1	III	57.9	29.6	51.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0949	NM_004223	Huesken	UBE2L6	9246	UGCGCAGGUUGAAGGCUUUca	AAAGCCUUCAACCUGCGCA	-0.9	-2.1	0.3	-1	-1	4	-40	45.1	-6.6	5	II	52.6	43.9	69.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0950	NM_004223	Huesken	UBE2L6	9246	GUAGGGAGGUUGGUCGGGUag	ACCCGACCAACCUCCCUAC	-2.2	-2.2	1.7	1	2	4	-44.3	29.9	-1.6	2	III	63.2	39.9	64	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0951	NM_004223	Huesken	UBE2L6	9246	GGGAGGUUGGUCGGGUAGGag	CCUACCCGACCAACCUCCC	-3.3	-3.3	1.7	1	0	4	-45.4	30.3	-2.4	1	II	68.4	37.8	55.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0952	NM_004223	Huesken	UBE2L6	9246	GAGGUUGGUCGGGUAGGAGga	CUCCUACCCGACCAACCUC	-2.1	-2.4	2	0	2	4	-43.3	41.1	2.3	2	II	63.2	38.3	63.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0953	NM_004223	Huesken	UBE2L6	9246	GGUUGGUCGGGUAGGAGGAga	UCCUCCUACCCGACCAACC	-2.4	-3.3	2	1	-1	4	-44.5	38.2	-3.3	2	III	63.2	28.1	49.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0954	NM_004223	Huesken	UBE2L6	9246	UUGGUCGGGUAGGAGGAGAgc	UCUCCUCCUACCCGACCAA	-2.4	-0.9	2	1	0	4	-43.5	46.1	-1.5	5	II	57.9	52.4	79.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0955	NM_004223	Huesken	UBE2L6	9246	GUCGGGUAGGAGGAGAGCGug	CGCUCUCCUCCUACCCGAC	-2.4	-2.2	2	2	2	4	-45.1	33	2.3	1	II	68.4	47.7	73.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0956	NM_004223	Huesken	UBE2L6	9246	GGUAGGAGGAGAGCGUGCCac	GGCACGCUCUCCUCCUACC	-3.3	-3.3	0.6	1	2	3	-45.8	37	7.4	3	II	68.4	44.9	62.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0957	NM_004223	Huesken	UBE2L6	9246	UAGGAGGAGAGCGUGCCACac	GUGGCACGCUCUCCUCCUA	-2.2	-1.3	0.6	1	4	3	-44.6	47.5	10.1	6	II	63.2	64.3	96.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0958	NM_004223	Huesken	UBE2L6	9246	AGGAGAGCGUGCCACACCAgg	UGGUGUGGCACGCUCUCCU	-2.1	-2.1	-2.1	2	0	3	-45.2	35.9	-3.7	2	II	63.2	42.9	89.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0959	NM_004223	Huesken	UBE2L6	9246	GAGCGUGCCACACCAGGACau	GUCCUGGUGUGGCACGCUC	-2.2	-2.4	-2.1	2	1	3	-45.3	36.6	7.5	1	II	68.4	41.7	74.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0960	NM_004223	Huesken	UBE2L6	9246	CAGGACAUUGGCAUCAUCGcu	CGAUGAUGCCAAUGUCCUG	-2.4	-2.1	0.1	-1	1	3	-38.6	61.8	-1.9	4	II	52.6	60.6	97.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0961	NM_004223	Huesken	UBE2L6	9246	AGGACAUUGGCAUCAUCGCug	GCGAUGAUGCCAAUGUCCU	-3.4	-2.1	0.1	3	3	3	-39.9	61.2	5.1	4	Ib	52.6	57.3	107.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0962	NM_004223	Huesken	UBE2L6	9246	GGACAUUGGCAUCAUCGCUgg	AGCGAUGAUGCCAAUGUCC	-2.1	-3.3	-0.1	3	0	3	-39.9	52.3	6.5	3	II	52.6	40.8	87.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0963	NM_004223	Huesken	UBE2L6	9246	GGCAUCAUCGCUGGACAGGuu	CCUGUCCAGCGAUGAUGCC	-3.3	-3.3	-0.1	0	0	3	-43.3	43.4	0.7	2	II	63.2	37.8	77.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0964	NM_004223	Huesken	UBE2L6	9246	CGCUGGACAGGUUCCGCAGgu	CUGCGGAACCUGUCCAGCG	-2.1	-2.4	-2.4	-1	-1	4	-44.2	45.4	1	0	II	68.4	37.3	76.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0965	NM_004223	Huesken	UBE2L6	9246	CUGGACAGGUUCCGCAGGUau	ACCUGCGGAACCUGUCCAG	-2.2	-2.1	-2.4	-1	-1	4	-43.9	31.7	-3.2	1	III	63.2	38.7	73.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0966	NM_004223	Huesken	UBE2L6	9246	ACAGGUUCCGCAGGUAUGGgg	CCAUACCUGCGGAACCUGU	-3.3	-2.2	-0.3	2	3	4	-41.8	44.6	-2.4	4	Ib	57.9	55.9	88.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0967	NM_016406	Huesken	Ufc1	51506	AUUUCUCUUUGUGUUGGAUga	AUCCAACACAAAGAGAAAU	-1.1	-1.1	2.4	1	3	2	-32.1	88.6	1.4	5	II	31.6	48.7	69.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0968	NM_016406	Huesken	Ufc1	51506	UUCUCUUUGUGUUGGAUGAcg	UCAUCCAACACAAAGAGAA	-2.4	-0.9	2.4	1	0	2	-34.6	91.1	-0.7	7	II	36.8	58.8	93	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0969	NM_016406	Huesken	Ufc1	51506	CUCUUUGUGUUGGAUGACGcc	CGUCAUCCAACACAAAGAG	-2.4	-2.1	1	0	0	2	-35.9	62.9	-2	4	II	47.4	54.2	86.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0970	NM_016406	Huesken	Ufc1	51506	UUGUGUUGGAUGACGCCCUuc	AGGGCGUCAUCCAACACAA	-2.1	-0.9	-3.4	2	2	5	-40.5	65.1	-6.6	6	II	52.6	71.8	107.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0971	NM_016406	Huesken	Ufc1	51506	CCCUUCUGAAUCAGAUCAGgg	CUGAUCUGAUUCAGAAGGG	-2.1	-3.3	-3.4	-1	-1	3	-37.3	56.3	-6.9	3	II	47.4	40.7	73.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0972	NM_016406	Huesken	Ufc1	51506	UGAAUCAGAUCAGGGAUUUcc	AAAUCCCUGAUCUGAUUCA	-0.9	-2.1	0.2	-1	0	3	-35.1	64.7	-6.3	8	II	36.8	55.2	81	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0973	NM_016406	Huesken	Ufc1	51506	UCAGGGAUUUCCACUGCCAgc	UGGCAGUGGAAAUCCCUGA	-2.1	-2.4	-0.6	0	1	3	-41.5	48	-11	5	II	52.6	56.7	71.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0974	NM_016406	Huesken	Ufc1	51506	AGGGAUUUCCACUGCCAGCca	GCUGGCAGUGGAAAUCCCU	-3.4	-2.1	-0.6	2	2	3	-42.5	57.9	10.1	3	Ib	57.9	58.4	48.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0975	NM_016406	Huesken	Ufc1	51506	ACUGCCAGCCAUGGACCCAgc	UGGGUCCAUGGCUGGCAGU	-2.1	-2.2	-1.3	1	1	3	-45.9	38.8	1.6	2	II	63.2	40.1	35.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0976	NM_016406	Huesken	Ufc1	51506	CUGCCAGCCAUGGACCCAGcc	CUGGGUCCAUGGCUGGCAG	-2.1	-2.1	-1.3	1	0	3	-45.8	33.3	-2	1	II	68.4	45.7	60.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0977	NM_016406	Huesken	Ufc1	51506	UGGACCCAGCCCCAGAGCCau	GGCUCUGGGGCUGGGUCCA	-3.3	-2.1	-4.6	1	3	5	-49.5	44.1	5.1	2	II	73.7	58	69.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0978	NM_016406	Huesken	Ufc1	51506	AGCCCCAGAGCCAUGAGAUga	AUCUCAUGGCUCUGGGGCU	-1.1	-2.1	-3.4	2	1	5	-44.1	47.6	-8.9	1	II	57.9	39.7	53.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0979	NM_016406	Huesken	Ufc1	51506	AGCCAUGAGAUGAGCUAGUcc	ACUAGCUCAUCUCAUGGCU	-2.2	-2.1	-1.7	2	0	3	-39.8	47	-3.9	4	II	47.4	47.1	53.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0980	NM_016406	Huesken	Ufc1	51506	CCAUGAGAUGAGCUAGUCCaa	GGACUAGCUCAUCUCAUGG	-3.3	-3.3	1.3	-1	1	2	-40	53.7	12.4	4	II	52.6	49.2	61.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0981	NM_016406	Huesken	Ufc1	51506	UCCAAAUUUGGGCACAUUCcu	GAAUGUGCCCAAAUUUGGA	-2.4	-2.4	-1.5	1	1	4	-35.4	74.7	12.5	6	Ia	42.1	67	67.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0982	NM_016406	Huesken	Ufc1	51506	AUUUGGGCACAUUCCUGGCcc	GCCAGGAAUGUGCCCAAAU	-3.4	-1.1	-0.8	1	5	4	-40	70.9	12.7	6	Ib	52.6	63	77.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0983	NM_016406	Huesken	Ufc1	51506	AUUCCUGGCCCACAAAGGUuu	ACCUUUGUGGGCCAGGAAU	-2.2	-1.1	-1.5	3	3	5	-41	53	-3.9	4	II	52.6	53.9	56.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0984	NM_016406	Huesken	Ufc1	51506	ACAAAGGUUUGAAAUGAUCcg	GAUCAUUUCAAACCUUUGU	-2.4	-2.2	0.4	1	2	2	-30.9	66	7.5	6	Ib	31.6	61.1	98.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0985	NM_016406	Huesken	Ufc1	51506	AAGGUUUGAAAUGAUCCGUca	ACGGAUCAUUUCAAACCUU	-2.2	-0.9	-0.3	2	2	3	-33.6	76	1.4	6	II	36.8	63.2	106	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0986	NM_016406	Huesken	Ufc1	51506	AGGUUUGAAAUGAUCCGUCag	GACGGAUCAUUUCAAACCU	-2.4	-2.1	-0.3	1	3	3	-35.1	71.1	7.1	6	Ib	42.1	63.9	122.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0987	NM_016406	Huesken	Ufc1	51506	GUUUGAAAUGAUCCGUCAGgc	CUGACGGAUCAUUUCAAAC	-2.1	-2.2	-0.3	0	2	3	-33.9	69.9	7.7	4	II	42.1	47.1	65.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0988	NM_016406	Huesken	Ufc1	51506	UUGAAAUGAUCCGUCAGGCau	GCCUGACGGAUCAUUUCAA	-3.4	-0.9	-0.4	1	5	3	-37.5	67.8	7.7	7	Ia	47.4	71.4	117.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0989	NM_016406	Huesken	Ufc1	51506	CCGUCAGGCAUAUUUUGCCac	GGCAAAAUAUGCCUGACGG	-3.3	-3.3	-4.7	0	1	3	-38.3	60.9	7.7	2	II	52.6	50.3	99.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0990	NM_016406	Huesken	Ufc1	51506	UGCCACCCCUGUACAUCUUug	AAGAUGUACAGGGGUGGCA	-0.9	-2.1	0.5	2	0	4	-41.5	56.3	-0.8	4	II	52.6	45.4	62.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0991	NM_016406	Huesken	Ufc1	51506	CACCCCUGUACAUCUUUGCug	GCAAAGAUGUACAGGGGUG	-3.4	-2.1	0.5	1	0	4	-39.1	68.9	3.1	4	II	52.6	53.3	108.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0992	NM_016406	Huesken	Ufc1	51506	CCUGUACAUCUUUGCUGUCuu	GACAGCAAAGAUGUACAGG	-2.4	-3.3	-0.1	-2	0	2	-37	54.4	8.4	2	II	47.4	39.6	50.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0993	NM_016406	Huesken	Ufc1	51506	UGUACAUCUUUGCUGUCUUuc	AAGACAGCAAAGAUGUACA	-0.9	-2.1	-0.1	1	2	2	-34.6	81.3	0.8	6	II	36.8	58.5	112.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0994	NM_016406	Huesken	Ufc1	51506	AUCUUUGCUGUCUUUCCAUcc	AUGGAAAGACAGCAAAGAU	-1.1	-1.1	0.3	1	2	2	-34.5	81.4	-4	6	II	36.8	57.4	69.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0995	NM_016406	Huesken	Ufc1	51506	CUUUGCUGUCUUUCCAUCCag	GGAUGGAAAGACAGCAAAG	-3.3	-2.1	0.3	0	1	2	-36.7	79.3	8.4	5	II	47.4	63.4	102.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0996	NM_016406	Huesken	Ufc1	51506	UCUUUCCAUCCAGCUCAGGaa	CCUGAGCUGGAUGGAAAGA	-3.3	-2.4	-1	0	3	2	-40.5	60.5	-2.4	5	Ib	52.6	57.6	120.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0997	NM_016406	Huesken	Ufc1	51506	CAUCCAGCUCAGGAACUGCaa	GCAGUUCCUGAGCUGGAUG	-3.4	-2.1	-4.4	0	1	2	-41.6	47.5	7.7	2	II	57.9	48.2	80.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0998	NM_016406	Huesken	Ufc1	51506	CCAGCUCAGGAACUGCAAUuu	AUUGCAGUUCCUGAGCUGG	-1.1	-3.3	-4.4	0	-2	2	-40.1	47.1	1.1	0	III	52.6	32.5	61.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si0999	NM_016406	Huesken	Ufc1	51506	GAACUGCAAUUUCUGGGGCag	GCCCCAGAAAUUGCAGUUC	-3.4	-2.4	0.6	1	4	5	-38.9	56.4	15.2	3	II	52.6	46.8	61.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1000	NM_016406	Huesken	Ufc1	51506	AACUGCAAUUUCUGGGGCAgu	UGCCCCAGAAAUUGCAGUU	-2.1	-0.9	-0.1	0	1	5	-38.6	56.7	-6.1	5	II	47.4	47.2	44.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1001	NM_016406	Huesken	Ufc1	51506	UGCAAUUUCUGGGGCAGUAgu	UACUGCCCCAGAAAUUGCA	-1.3	-2.1	-0.1	1	-1	5	-39	60.2	-1.1	6	II	47.4	48.1	52.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1002	NM_016406	Huesken	Ufc1	51506	UGGGGCAGUAGUAGGAUAUgu	AUAUCCUACUACUGCCCCA	-1.1	-2.1	3.5	-1	-1	5	-40	58.8	-3.5	6	II	47.4	51.6	78.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1003	NM_016406	Huesken	Ufc1	51506	CAGUAGUAGGAUAUGUGAUag	AUCACAUAUCCUACUACUG	-1.1	-2.1	3.5	1	-1	2	-34.5	71	1.8	4	II	36.8	44.3	69.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1004	NM_016406	Huesken	Ufc1	51506	AUAUGUGAUAGGAAUGUCAaa	UGACAUUCCUAUCACAUAU	-2.1	-1.1	1	1	1	2	-33.3	75.7	-6.3	8	II	31.6	63.7	114.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1005	NM_016406	Huesken	Ufc1	51506	GAUAGGAAUGUCAAACUCGau	CGAGUUUGACAUUCCUAUC	-2.4	-2.4	0	0	2	2	-34.3	58.2	-4.9	5	II	42.1	60.5	95.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1006	NM_016406	Huesken	Ufc1	51506	UCAAACUCGAUGUCAAACUca	AGUUUGACAUCGAGUUUGA	-2.1	-2.4	0.8	2	1	2	-33.8	77.2	-1.3	8	II	36.8	69.7	121.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1007	NM_016406	Huesken	Ufc1	51506	GAUGUCAAACUCAUAUUUCag	GAAAUAUGAGUUUGACAUC	-2.4	-2.4	1.3	0	2	1	-30.6	77.6	4.8	5	II	31.6	56.4	66.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1008	NM_016406	Huesken	Ufc1	51506	AUGUCAAACUCAUAUUUCAgg	UGAAAUAUGAGUUUGACAU	-2.1	-1.1	1.3	2	0	1	-30.3	82.2	-6.1	6	II	26.3	57.2	75.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1009	NM_016406	Huesken	Ufc1	51506	AACUCAUAUUUCAGGAGGUca	ACCUCCUGAAAUAUGAGUU	-2.2	-0.9	0.2	1	2	2	-34.9	66.1	-8.9	5	II	36.8	51.8	43.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1010	NM_016406	Huesken	Ufc1	51506	CAGGAGGUCAUGGAUAUACca	GUAUAUCCAUGACCUCCUG	-2.2	-2.1	2.7	0	-1	2	-37.9	54.2	8.1	5	II	47.4	53.5	96	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1011	NM_016406	Huesken	Ufc1	51506	GGAGGUCAUGGAUAUACCAgc	UGGUAUAUCCAUGACCUCC	-2.1	-3.3	-1.4	0	-1	2	-39.1	47	-3.8	0	III	47.4	37.2	31.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1012	NM_016406	Huesken	Ufc1	51506	AUAUACCAGCAUUUUCCAAac	UUGGAAAAUGCUGGUAUAU	-0.9	-1.1	1.5	2	2	2	-32.5	80.6	6.6	6	II	31.6	48.6	56.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1013	NM_016406	Huesken	Ufc1	51506	UUUCCAAACCACCGAGUUCcu	GAACUCGGUGGUUUGGAAA	-2.4	-0.9	-1.1	1	3	3	-36.9	67.6	9.8	6	Ia	47.4	65.4	97.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1014	NM_016406	Huesken	Ufc1	51506	CCAAACCACCGAGUUCCUUcc	AAGGAACUCGGUGGUUUGG	-0.9	-3.3	-1.5	0	-1	3	-39	44.7	-0.9	2	III	52.6	34.5	32.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1015	NM_016406	Huesken	Ufc1	51506	AACCACCGAGUUCCUUCCUug	AGGAAGGAACUCGGUGGUU	-2.1	-0.9	0.5	2	2	3	-40.5	56.7	4.2	4	II	52.6	49	61.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1016	NM_016406	Huesken	Ufc1	51506	GUUCCUUCCUUGUUGGACUcc	AGUCCAACAAGGAAGGAAC	-2.1	-2.2	-2.1	3	1	2	-37.8	73.8	-1.6	2	II	47.4	50	83.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1017	NM_016406	Huesken	Ufc1	51506	CUUCCUUGUUGGACUCCAGuc	CUGGAGUCCAACAAGGAAG	-2.1	-2.1	-2.1	1	1	2	-38.9	61.5	-4.6	2	II	52.6	48.3	66.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1018	NM_016406	Huesken	Ufc1	51506	UGGACUCCAGUCGGAACCAau	UGGUUCCGACUGGAGUCCA	-2.1	-2.1	-2.8	-1	0	3	-43.2	44.5	-6.1	2	II	57.9	48.9	71.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1019	NM_016406	Huesken	Ufc1	51506	UCGGAACCAAUCGUUGUCAgc	UGACAACGAUUGGUUCCGA	-2.1	-2.4	0.1	0	0	3	-37.7	55.5	-1.1	6	II	47.4	52.2	76.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1020	NM_016406	Huesken	Ufc1	51506	GGAACCAAUCGUUGUCAGCau	GCUGACAACGAUUGGUUCC	-3.4	-3.3	0.1	-1	1	2	-38.4	60.1	10.4	3	II	52.6	43.6	78	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1021	NM_016406	Huesken	Ufc1	51506	CGUUGUCAGCAUUCUUGUUgu	AACAAGAAUGCUGACAACG	-0.9	-2.4	1.7	0	-2	2	-34.5	69.2	-0.7	2	III	42.1	34.2	43.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1022	NM_016406	Huesken	Ufc1	51506	UUGUCAGCAUUCUUGUUGUuc	ACAACAAGAAUGCUGACAA	-2.2	-0.9	1.7	0	2	2	-34.2	84.9	-4	7	II	36.8	63.1	86.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1023	NM_016406	Huesken	Ufc1	51506	CAUUCUUGUUGUUCUCCACau	GUGGAGAACAACAAGAAUG	-2.2	-2.1	0.3	0	1	2	-34.4	86	8.4	4	II	42.1	52.4	68.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1024	NM_016406	Huesken	Ufc1	51506	GUUGUUCUCCACAUACCGGau	CCGGUAUGUGGAGAACAAC	-3.3	-2.2	-0.7	1	4	4	-38.5	63.3	-0.3	3	II	52.6	52.9	72.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1025	NM_016406	Huesken	Ufc1	51506	UCUCCACAUACCGGAUAAGgg	CUUAUCCGGUAUGUGGAGA	-2.1	-2.4	-2.4	0	2	4	-38	52.4	-4.7	4	II	47.4	48.3	78.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1026	NM_016406	Huesken	Ufc1	51506	ACAUACCGGAUAAGGGACUga	AGUCCCUUAUCCGGUAUGU	-2.1	-2.2	-0.2	1	1	4	-39	47	-3.9	6	II	47.4	48.4	74.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1027	NM_016406	Huesken	Ufc1	51506	CAUACCGGAUAAGGGACUGau	CAGUCCCUUAUCCGGUAUG	-2.1	-2.1	-0.2	-2	1	4	-38.9	50.4	0.3	4	II	52.6	43.9	78.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1028	NM_016406	Huesken	Ufc1	51506	AUACCGGAUAAGGGACUGAua	UCAGUCCCUUAUCCGGUAU	-2.4	-1.1	-0.2	1	1	4	-39.2	48.2	-1.5	7	II	47.4	49.7	88.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1029	NM_016406	Huesken	Ufc1	51506	UACCGGAUAAGGGACUGAUau	AUCAGUCCCUUAUCCGGUA	-1.1	-1.3	-0.2	1	1	4	-39.2	58.3	-4	6	II	47.4	55.5	99.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1030	NM_016406	Huesken	Ufc1	51506	CUGAUAUUCCUCCUUCAGUcg	ACUGAAGGAGGAAUAUCAG	-2.2	-2.1	-0.9	1	-1	2	-36.3	72.8	-3.6	4	II	42.1	52	75.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1031	NM_016406	Huesken	Ufc1	51506	AUAUUCCUCCUUCAGUCGCug	GCGACUGAAGGAGGAAUAU	-3.4	-1.1	0.9	2	5	3	-37.9	76	12.8	6	Ia	47.4	66.3	106.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1032	NM_016406	Huesken	Ufc1	51506	UAUUCCUCCUUCAGUCGCUgc	AGCGACUGAAGGAGGAAUA	-2.1	-1.3	0.9	1	3	3	-38.9	72.5	-8.9	7	II	47.4	63.9	93.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1033	NM_016406	Huesken	Ufc1	51506	AUUCCUCCUUCAGUCGCUGca	CAGCGACUGAAGGAGGAAU	-2.1	-1.1	0.4	2	5	3	-39.7	59.7	0	4	Ib	52.6	57	100.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1034	NM_016406	Huesken	Ufc1	51506	UCCUUCAGUCGCUGCACCCac	GGGUGCAGCGACUGAAGGA	-3.3	-2.4	-0.2	-1	3	3	-44.2	59.4	5.1	5	Ib	63.2	62.9	85.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1035	NM_016406	Huesken	Ufc1	51506	AGUCGCUGCACCCACAACUca	AGUUGUGGGUGCAGCGACU	-2.1	-2.1	-4.7	4	1	3	-42.6	53.7	-6.3	3	II	57.9	55.3	80.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1036	NM_016406	Huesken	Ufc1	51506	UCACGAUCUCGGGGUCCGGcg	CCGGACCCCGAGAUCGUGA	-3.3	-2.4	-1.4	1	4	5	-45.4	35	0	3	Ib	68.4	53.9	56.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1037	NM_014176	Huesken	HSPC150	29089	UUUCUUUUCUAUGCCUACUag	AGUAGGCAUAGAAAAGAAA	-2.1	-0.9	2.7	3	2	3	-32.4	96.4	1.4	7	II	31.6	73.6	101.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1038	NM_014176	Huesken	HSPC150	29089	CUUUUCUAUGCCUACUAGCug	GCUAGUAGGCAUAGAAAAG	-3.4	-2.1	-2.1	0	1	3	-35	80.2	0.7	4	II	42.1	62	106	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1039	NM_014176	Huesken	HSPC150	29089	UACUAGCUGACUGGCCUUCcu	GAAGGCCAGUCAGCUAGUA	-2.4	-1.3	-1.6	0	2	4	-40.8	63	10.8	7	Ib	52.6	62.4	91.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1040	NM_014176	Huesken	HSPC150	29089	CCUUCCUUUUCUGUGUUGAgu	UCAACACAGAAAAGGAAGG	-2.4	-3.3	1.2	0	-2	2	-35.3	75.7	-2.8	4	III	42.1	36.1	65.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1041	NM_014176	Huesken	HSPC150	29089	UGUUGAGUUGUGUACUCUGga	CAGAGUACACAACUCAACA	-2.1	-2.1	-3.5	0	3	1	-35.7	70.7	0	5	Ib	42.1	65.5	83	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1042	NM_014176	Huesken	HSPC150	29089	UUGUGUACUCUGGAGUCACca	GUGACUCCAGAGUACACAA	-2.2	-0.9	-0.5	0	3	2	-38.4	66.4	7.4	6	Ia	47.4	69.3	108	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1043	NM_014176	Huesken	HSPC150	29089	GUGUACUCUGGAGUCACCAgc	UGGUGACUCCAGAGUACAC	-2.1	-2.2	-1.6	1	0	2	-40.8	55.9	1.3	3	II	52.6	45.1	80.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1044	NM_014176	Huesken	HSPC150	29089	GUCACCAGCCUCUGGUAGAuu	UCUACCAGAGGCUGGUGAC	-2.4	-2.2	-5.4	0	-2	3	-43.1	49.1	-6.1	1	III	57.9	33.6	63.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1045	NM_014176	Huesken	HSPC150	29089	UCACCAGCCUCUGGUAGAUua	AUCUACCAGAGGCUGGUGA	-1.1	-2.4	-5.4	1	1	3	-42	46.3	-3.3	4	II	52.6	41.7	97	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1046	NM_014176	Huesken	HSPC150	29089	CUCUGGUAGAUUAUCAAGCau	GCUUGAUAAUCUACCAGAG	-3.4	-2.1	0.9	1	1	2	-35.3	73	8.2	4	II	42.1	63.1	101	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1047	NM_014176	Huesken	HSPC150	29089	AGAUUAUCAAGCAUCUCUUcc	AAGAGAUGCUUGAUAAUCU	-0.9	-2.1	-1.3	2	2	2	-32.9	82.6	-6.6	6	II	31.6	58.8	99.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1048	NM_014176	Huesken	HSPC150	29089	GAUUAUCAAGCAUCUCUUCcu	GAAGAGAUGCUUGAUAAUC	-2.4	-2.4	0.5	0	1	2	-33.2	76.1	7.8	5	II	36.8	50.6	85.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1049	NM_014176	Huesken	HSPC150	29089	UUAUCAAGCAUCUCUUCCUca	AGGAAGAGAUGCUUGAUAA	-2.1	-0.9	0.7	1	3	2	-35.1	85.3	-6.3	8	II	36.8	74.5	113	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1050	NM_014176	Huesken	HSPC150	29089	AUCAAGCAUCUCUUCCUCAuc	UGAGGAAGAGAUGCUUGAU	-2.1	-1.1	0.7	1	1	2	-37.4	67.3	-1.1	6	II	42.1	55.9	91.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1051	NM_014176	Huesken	HSPC150	29089	CAAGCAUCUCUUCCUCAUCag	GAUGAGGAAGAGAUGCUUG	-2.4	-2.1	0.7	0	0	2	-37.4	65.4	10.8	3	II	47.4	54.8	95.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1052	NM_014176	Huesken	HSPC150	29089	GCAUCUCUUCCUCAUCAGCcu	GCUGAUGAGGAAGAGAUGC	-3.4	-3.4	0.8	0	1	2	-39.9	62	10.5	2	II	52.6	43.6	73.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1053	NM_014176	Huesken	HSPC150	29089	UUUCUGUCUUGCAUGCUUCuc	GAAGCAUGCAAGACAGAAA	-2.4	-0.9	1.8	2	4	2	-35.6	84.1	4.8	8	Ia	42.1	75.9	97.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1054	NM_014176	Huesken	HSPC150	29089	CUGUCCACUGUCUGGCAUUcu	AAUGCCAGACAGUGGACAG	-0.9	-2.1	-1.3	-2	-3	3	-40.2	55.7	-8.3	2	III	52.6	42.6	77.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1055	NM_014176	Huesken	HSPC150	29089	GUCCACUGUCUGGCAUUCUug	AGAAUGCCAGACAGUGGAC	-2.1	-2.2	-1.3	2	-1	3	-40.5	49.8	-1.3	3	III	52.6	41.1	95	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1056	NM_014176	Huesken	HSPC150	29089	UCUGGCAUUCUUGAGGAAGgc	CUUCCUCAAGAAUGCCAGA	-2.1	-2.4	1.1	0	2	3	-38.1	60.9	3	5	II	47.4	59.7	76	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1057	NM_014176	Huesken	HSPC150	29089	AUUAUAUUUAAAUUCUGAGga	CUCAGAAUUUAAAUAUAAU	-2.1	-1.1	2.8	2	5	1	-24.8	108.8	4.6	8	Ia	15.8	73.1	85.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1058	NM_014176	Huesken	HSPC150	29089	UUAAAUUCUGAGGAUAUGUca	ACAUAUCCUCAGAAUUUAA	-2.2	-0.9	1.6	1	2	2	-30.6	89.1	1	9	II	26.3	76.3	114	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1059	NM_014176	Huesken	HSPC150	29089	AGCCAUGAGCGGGUCAUCAgg	UGAUGACCCGCUCAUGGCU	-2.1	-2.1	-1.1	3	0	5	-43.3	45.5	1.3	3	II	57.9	43.4	75.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1060	NM_014176	Huesken	HSPC150	29089	CCAUGAGCGGGUCAUCAGGgu	CCUGAUGACCCGCUCAUGG	-3.3	-3.3	-0.8	0	0	5	-43.2	40.9	3.4	2	II	63.2	41.7	83.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1061	NM_014176	Huesken	HSPC150	29089	CAUGAGCGGGUCAUCAGGGuu	CCCUGAUGACCCGCUCAUG	-3.3	-2.1	-0.8	0	3	5	-43.2	38.6	-1.9	2	II	63.2	50.2	91.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1062	NM_014176	Huesken	HSPC150	29089	AUGAGCGGGUCAUCAGGGUug	ACCCUGAUGACCCGCUCAU	-2.2	-1.1	-0.1	2	2	5	-43.3	44.4	-6.3	4	II	57.9	47.7	100.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1063	NM_014176	Huesken	HSPC150	29089	GGGUCAUCAGGGUUGGGUUcu	AACCCAACCCUGAUGACCC	-0.9	-3.3	1.6	1	-1	3	-42.5	55.9	-1.6	1	III	57.9	30.6	52.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1064	NM_014176	Huesken	HSPC150	29089	UGGGUUCUGACAUGAGCAGcu	CUGCUCAUGUCAGAACCCA	-2.1	-2.1	0.1	0	2	3	-40.4	56.2	-2.4	5	II	52.6	54.8	104.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1065	NM_014176	Huesken	HSPC150	29089	GGUUCUGACAUGAGCAGCUga	AGCUGCUCAUGUCAGAACC	-2.1	-3.3	-0.1	0	0	2	-40.5	52.3	-6.6	4	III	52.6	42	79.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1066	NM_014176	Huesken	HSPC150	29089	GCAGCUGAAUAGAGGUCAAca	UUGACCUCUAUUCAGCUGC	-0.9	-3.4	0.6	-1	-2	2	-38.4	41.6	-4	2	III	47.4	23.4	49.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1067	NM_014176	Huesken	HSPC150	29089	UAGAGGUCAACACAGUUGCga	GCAACUGUGUUGACCUCUA	-3.4	-1.3	-0.8	1	3	2	-38.1	72.7	7.5	8	Ib	47.4	75.9	114.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1068	NM_014176	Huesken	HSPC150	29089	GGUCAACACAGUUGCGAUGuu	CAUCGCAACUGUGUUGACC	-2.1	-3.3	-0.8	1	0	3	-38.2	51.4	5.7	3	II	52.6	35.5	84.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1069	NM_014176	Huesken	HSPC150	29089	AACACAGUUGCGAUGUUGAgg	UCAACAUCGCAACUGUGUU	-2.4	-0.9	-0.8	1	1	3	-35.8	68.5	-6.4	6	II	42.1	50.5	102.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1070	NM_014176	Huesken	HSPC150	29089	ACAGUUGCGAUGUUGAGGGau	CCCUCAACAUCGCAACUGU	-3.3	-2.2	1.5	2	5	3	-39.3	62.1	2.3	6	Ib	52.6	63.6	106.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1071	NM_014176	Huesken	HSPC150	29089	GGGAUGGUCUCCAAGCACCuu	GGUGCUUGGAGACCAUCCC	-3.3	-3.3	-2.3	1	0	3	-44	43.6	0.1	1	II	63.2	48.1	80.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1072	NM_014176	Huesken	HSPC150	29089	GGAUGGUCUCCAAGCACCUuu	AGGUGCUUGGAGACCAUCC	-2.1	-3.3	-2.5	0	0	2	-42.8	37.7	-8.9	1	III	57.9	40.2	65.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1073	NM_014176	Huesken	HSPC150	29089	GAUGGUCUCCAAGCACCUUuu	AAGGUGCUUGGAGACCAUC	-0.9	-2.4	-2.5	1	0	2	-40.4	46.5	-1.7	1	III	52.6	37.3	60.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1074	NM_014176	Huesken	HSPC150	29089	GUCUCCAAGCACCUUUUGGug	CCAAAAGGUGCUUGGAGAC	-3.3	-2.2	-2	1	2	2	-38.7	70.3	4.7	3	II	52.6	47.1	84.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1075	NM_014176	Huesken	HSPC150	29089	AGCACCUUUUGGUGGCAAUuu	AUUGCCACCAAAAGGUGCU	-1.1	-2.1	-4.2	1	-1	3	-38.4	65.8	-4	3	II	47.4	38.5	43.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1076	NM_014176	Huesken	HSPC150	29089	CUUUUGGUGGCAAUUUGAGaa	CUCAAAUUGCCACCAAAAG	-2.1	-2.1	1.3	0	2	3	-33.7	68.2	-4.6	5	II	42.1	51	68.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1077	NM_014176	Huesken	HSPC150	29089	UGGUGGCAAUUUGAGAACAuc	UGUUCUCAAAUUGCCACCA	-2.1	-2.1	1.7	0	-1	3	-36.5	57.3	-0.7	4	II	42.1	51	53.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1078	NM_014176	Huesken	HSPC150	29089	GGCAAUUUGAGAACAUCCAga	UGGAUGUUCUCAAAUUGCC	-2.1	-3.3	1	2	-2	3	-35.7	61.8	-3.6	4	II	42.1	41.7	71.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1079	NM_014176	Huesken	HSPC150	29089	CAAUUUGAGAACAUCCAGAca	UCUGGAUGUUCUCAAAUUG	-2.4	-2.1	0.9	0	-1	2	-33.5	72.4	-6.1	6	II	36.8	53.5	80.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1080	NM_014176	Huesken	HSPC150	29089	GAGAACAUCCAGACAAAUCcu	GAUUUGUCUGGAUGUUCUC	-2.4	-2.4	1.4	1	0	2	-35.1	66.5	9.8	4	II	42.1	53.9	75	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1081	NM_014176	Huesken	HSPC150	29089	AACAUCCAGACAAAUCCUUcc	AAGGAUUUGUCUGGAUGUU	-0.9	-0.9	0	2	1	2	-34.5	65.6	-8.9	6	II	36.8	54.9	79.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1082	NM_014176	Huesken	HSPC150	29089	CAUCCAGACAAAUCCUUCCag	GGAAGGAUUUGUCUGGAUG	-3.3	-2.1	0	1	1	2	-37.1	71.1	7.7	5	II	47.4	59.6	96.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1083	NM_014176	Huesken	HSPC150	29089	GACAAAUCCUUCCAGCAGAau	UCUGCUGGAAGGAUUUGUC	-2.4	-2.4	0.1	2	-2	2	-38.2	53.6	-1	4	II	47.4	38.8	87.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1084	NM_014176	Huesken	HSPC150	29089	AAUCCUUCCAGCAGAAUCAau	UGAUUCUGCUGGAAGGAUU	-2.1	-0.9	-3	3	1	2	-37.1	72.3	-8.7	6	II	42.1	59.1	95.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1085	NM_014176	Huesken	HSPC150	29089	AAUCAAUGUUUGGAUGAUAaa	UAUCAUCCAAACAUUGAUU	-1.3	-0.9	-0.1	3	0	2	-30.4	66.2	-1.1	6	II	26.3	50.3	67.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1086	NM_014176	Huesken	HSPC150	29089	UGUUUGGAUGAUAAAUUGGag	CCAAUUUAUCAUCCAAACA	-3.3	-2.1	3.5	0	3	2	-31.1	84.4	2.7	8	Ib	31.6	70.8	101.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1087	NM_014176	Huesken	HSPC150	29089	UUUGGAUGAUAAAUUGGAGug	CUCCAAUUUAUCAUCCAAA	-2.1	-0.9	4	1	5	2	-31.3	78.6	2	6	Ia	31.6	66.5	107.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1088	NM_014176	Huesken	HSPC150	29089	GGAUGAUAAAUUGGAGUGAga	UCACUCCAAUUUAUCAUCC	-2.4	-3.3	4.5	0	-2	2	-34.1	55.1	-0.7	5	II	36.8	33.8	56.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1089	NM_014176	Huesken	HSPC150	29089	AAAUUGGAGUGAGAAAUCGga	CGAUUUCUCACUCCAAUUU	-2.4	-0.9	0.9	2	4	2	-32.6	66.2	0	8	Ia	36.8	67.2	106.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1090	NM_014176	Huesken	HSPC150	29089	AUUGGAGUGAGAAAUCGGAuc	UCCGAUUUCUCACUCCAAU	-2.4	-1.1	0.9	2	2	3	-36.5	54.2	-6.3	6	II	42.1	53.4	109.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1091	NM_014176	Huesken	HSPC150	29089	AAAUCGGAUCUGAGGAGGUuc	ACCUCCUCAGAUCCGAUUU	-2.2	-0.9	1.3	0	3	3	-38.9	51.4	-6.6	6	II	47.4	53.3	100.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1092	NM_014176	Huesken	HSPC150	29089	UCGGAUCUGAGGAGGUUCAaa	UGAACCUCCUCAGAUCCGA	-2.1	-2.4	0.1	0	-1	3	-41.4	52.7	-3.6	6	II	52.6	49.9	105.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1093	NM_014176	Huesken	HSPC150	29089	CGGAUCUGAGGAGGUUCAAau	UUGAACCUCCUCAGAUCCG	-0.9	-2.4	-0.5	-1	-4	3	-39.9	43.3	-5.8	3	III	52.6	25.2	56.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1094	NM_014176	Huesken	HSPC150	29089	GAGGUUCAAAUGGGUACCUcu	AGGUACCCAUUUGAACCUC	-2.1	-2.4	-1.2	-1	0	3	-38.1	49.8	-4	2	III	47.4	42.5	63.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1095	NM_014176	Huesken	HSPC150	29089	UUCAAAUGGGUACCUCUCAgg	UGAGAGGUACCCAUUUGAA	-2.1	-0.9	0.5	2	0	3	-37.1	67.2	1.4	8	II	42.1	63.2	87.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1096	NM_014176	Huesken	HSPC150	29089	CAAAUGGGUACCUCUCAGGaa	CCUGAGAGGUACCCAUUUG	-3.3	-2.1	-0.7	-1	1	3	-39.2	60.6	-6.7	6	II	52.6	58.3	78.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1097	NM_014176	Huesken	HSPC150	29089	AUGGGUACCUCUCAGGAAUga	AUUCCUGAGAGGUACCCAU	-1.1	-1.1	-0.7	2	0	3	-39.7	60.1	-0.9	4	II	47.4	51.3	114.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1098	NM_014176	Huesken	HSPC150	29089	ACCUCUCAGGAAUGAUAACuu	GUUAUCAUUCCUGAGAGGU	-2.2	-2.2	-0.7	1	1	2	-36.4	63.9	9.7	3	II	42.1	43.9	75.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1099	NM_014176	Huesken	HSPC150	29089	CUCUCAGGAAUGAUAACUUcu	AAGUUAUCAUUCCUGAGAG	-0.9	-2.1	0.2	-1	-1	2	-33.9	65.5	-6.2	4	III	36.8	46.6	79.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1100	NM_014176	Huesken	HSPC150	29089	AUGAUAACUUCUAGCUUAAaa	UUAAGCUAGAAGUUAUCAU	-0.9	-1.1	0.3	0	-1	2	-30.6	75.6	-5.7	7	II	26.3	49.3	89.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1101	NM_014176	Huesken	HSPC150	29089	AAAACACCUUUCUCAUAAGgu	CUUAUGAGAAAGGUGUUUU	-2.1	-0.9	0.6	1	3	2	-30.8	76.8	-2.4	5	Ia	31.6	56.9	92.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1102	NM_014176	Huesken	HSPC150	29089	CCUUUCUCAUAAGGUGUGUug	ACACACCUUAUGAGAAAGG	-2.2	-3.3	-1.9	-1	-2	2	-35.7	63.8	-3.7	4	II	42.1	35.9	74.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1103	NM_014176	Huesken	HSPC150	29089	GUGUUGGCUCCACCUAAUAuu	UAUUAGGUGGAGCCAACAC	-1.3	-2.2	-1.8	0	-3	3	-38.3	54.2	-3.7	3	III	47.4	37.3	57.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1104	NM_014176	Huesken	HSPC150	29089	CUCCACCUAAUAUUUGAGCuc	GCUCAAAUAUUAGGUGGAG	-3.4	-2.1	0.6	0	1	2	-35	71.3	10.4	3	II	42.1	53.7	87.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1105	NM_014176	Huesken	HSPC150	29089	CACCUAAUAUUUGAGCUCGca	CGAGCUCAAAUAUUAGGUG	-2.4	-2.1	1.1	0	0	2	-34.1	65.2	1	4	II	42.1	54.3	90	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1106	NM_014176	Huesken	HSPC150	29089	ACCUAAUAUUUGAGCUCGCag	GCGAGCUCAAAUAUUAGGU	-3.4	-2.2	1.8	1	3	3	-35.4	58.2	7.5	5	Ia	42.1	54	71.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1107	NM_014176	Huesken	HSPC150	29089	UAAUAUUUGAGCUCGCAGGuc	CCUGCGAGCUCAAAUAUUA	-3.3	-1.3	2.4	2	4	3	-35.3	89.1	0.4	9	Ia	42.1	79.9	102.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1108	NM_014176	Huesken	HSPC150	29089	AUAUUUGAGCUCGCAGGUCau	GACCUGCGAGCUCAAAUAU	-2.4	-1.1	2.1	2	4	3	-37.7	60.4	5.1	7	Ia	47.4	64.1	100.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1109	NM_014176	Huesken	HSPC150	29089	UCGCAGGUCAUCCAUUUGGuc	CCAAAUGGAUGACCUGCGA	-3.3	-2.4	-0.6	2	2	3	-39.6	59.4	2.8	7	II	52.6	66.1	95.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1110	NM_014176	Huesken	HSPC150	29089	GUCAUCCAUUUGGUCUUUAuc	UAAAGACCAAAUGGAUGAC	-1.3	-2.2	-1.1	0	-2	2	-33.7	66.3	-1.4	3	III	36.8	35.4	63.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1111	NM_014176	Huesken	HSPC150	29089	GGUCUUUAUCUUGCCAACAug	UGUUGGCAAGAUAAAGACC	-2.1	-3.3	0.8	2	-2	3	-35.7	63.5	-0.7	3	II	42.1	42.6	75.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1112	NM_014176	Huesken	HSPC150	29089	CUUGCCAACAUGUGAUGCCug	GGCAUCACAUGUUGGCAAG	-3.3	-2.1	-0.2	-1	2	3	-38.9	60.3	5.4	4	II	52.6	56.8	87.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1113	NM_014176	Huesken	HSPC150	29089	UGCCAACAUGUGAUGCCUGgg	CAGGCAUCACAUGUUGGCA	-2.1	-2.1	-0.5	1	3	3	-40.1	50.1	2.4	4	II	52.6	56.7	90.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1114	NM_001001481	Huesken	FLJ11011	55284	ACCAUUUUGUUUUCUUUGGau	CCAAAGAAAACAAAAUGGU	-3.3	-2.2	0.4	3	2	2	-30.1	94	3	6	Ia	31.6	62.7	101.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1115	NM_001001481	Huesken	FLJ11011	55284	UUUGUUUUCUUUGGAUUCUug	AGAAUCCAAAGAAAACAAA	-2.1	-0.9	1.6	1	2	2	-29.4	92.1	-1	8	II	26.3	71.4	105.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1116	NM_001001481	Huesken	FLJ11011	55284	UUUCUUUGGAUUCUUGUUAca	UAACAAGAAUCCAAAGAAA	-1.3	-0.9	1.6	3	1	2	-29.8	98.7	4	9	II	26.3	69.6	114.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1117	NM_001001481	Huesken	FLJ11011	55284	UUCUUGUUACAUGUUCGCAca	UGCGAACAUGUAACAAGAA	-2.1	-0.9	0.5	1	2	3	-33.7	87.8	3.9	7	II	36.8	64.3	103.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1118	NM_001001481	Huesken	FLJ11011	55284	UCUUGUUACAUGUUCGCACau	GUGCGAACAUGUAACAAGA	-2.2	-2.4	0.3	2	4	3	-35	85.3	9.7	7	Ia	42.1	71.5	125.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1119	NM_001001481	Huesken	FLJ11011	55284	UUGUUACAUGUUCGCACAUaa	AUGUGCGAACAUGUAACAA	-1.1	-0.9	-1.4	-1	1	3	-33.7	70.6	1.5	5	II	36.8	56.1	88.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1120	NM_001001481	Huesken	FLJ11011	55284	AAAAGAAUUAUCCGGUGGUcg	ACCACCGGAUAAUUCUUUU	-2.2	-0.9	0.1	0	3	4	-33.7	65.6	-3.9	7	II	36.8	58.8	96.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1121	NM_001001481	Huesken	FLJ11011	55284	AAAGAAUUAUCCGGUGGUCgu	GACCACCGGAUAAUUCUUU	-2.4	-0.9	0.1	1	4	4	-35.2	57.6	5.4	6	Ia	42.1	56.7	96.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1122	NM_001001481	Huesken	FLJ11011	55284	UAUCCGGUGGUCGUCUCUUuu	AAGAGACGACCACCGGAUA	-0.9	-1.3	-0.1	2	3	4	-40.5	59.1	-4	6	II	52.6	61	82.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1123	NM_001001481	Huesken	FLJ11011	55284	UGGUCGUCUCUUUUCCUUGca	CAAGGAAAAGAGACGACCA	-2.1	-2.1	0.3	1	2	2	-37.1	82.6	5.4	6	II	47.4	68	109.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1124	NM_001001481	Huesken	FLJ11011	55284	GGUCGUCUCUUUUCCUUGCag	GCAAGGAAAAGAGACGACC	-3.4	-3.3	2	2	1	2	-38.4	65.2	10.4	2	II	52.6	44	112.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1125	NM_001001481	Huesken	FLJ11011	55284	UUUCCUUGCAGCUGGAAAGca	CUUUCCAGCUGCAAGGAAA	-2.1	-0.9	-3	2	3	2	-37.4	79.8	-2.4	6	Ia	47.4	68.8	103	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1126	NM_001001481	Huesken	FLJ11011	55284	UCCUUGCAGCUGGAAAGCAug	UGCUUUCCAGCUGCAAGGA	-2.1	-2.4	-1.4	1	0	2	-41.1	44.5	-3.8	4	II	52.6	47.3	110.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1127	NM_001001481	Huesken	FLJ11011	55284	AGCUGGAAAGCAUGCUAAUaa	AUUAGCAUGCUUUCCAGCU	-1.1	-2.1	-1.2	0	-1	2	-36.8	54	-6.3	3	II	42.1	39.8	67.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1128	NM_001001481	Huesken	FLJ11011	55284	GGAAAGCAUGCUAAUAAUGcu	CAUUAUUAGCAUGCUUUCC	-2.1	-3.3	1.3	0	0	2	-32.5	58.5	0.8	3	II	36.8	46.6	93.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1129	NM_001001481	Huesken	FLJ11011	55284	AAAGCAUGCUAAUAAUGCUaa	AGCAUUAUUAGCAUGCUUU	-2.1	-0.9	-2.5	3	3	2	-32.3	73.8	-1.7	4	II	31.6	60.5	84.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1130	NM_001001481	Huesken	FLJ11011	55284	AAUGCUAAGACAAACUGAUug	AUCAGUUUGUCUUAGCAUU	-1.1	-0.9	-0.4	2	2	2	-32.3	71.1	-8.9	6	II	31.6	58.9	107	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1131	NM_001001481	Huesken	FLJ11011	55284	AUGCUAAGACAAACUGAUUgg	AAUCAGUUUGUCUUAGCAU	-0.9	-1.1	-0.4	2	0	2	-32.3	67.5	-4.3	5	II	31.6	55	104.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1132	NM_001001481	Huesken	FLJ11011	55284	CUAAGACAAACUGAUUGGAcu	UCCAAUCAGUUUGUCUUAG	-2.4	-2.1	0.6	-1	-1	2	-33.5	57.9	-5.1	4	II	36.8	41	85	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1133	NM_001001481	Huesken	FLJ11011	55284	AGACAAACUGAUUGGACUGag	CAGUCCAAUCAGUUUGUCU	-2.1	-2.1	0.6	1	3	2	-35.6	61.9	8	5	Ia	42.1	56.4	121.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1134	NM_001001481	Huesken	FLJ11011	55284	CUGAUUGGACUGAGAGCGCug	GCGCUCUCAGUCCAAUCAG	-3.4	-2.1	1.5	-2	1	4	-41	44.4	2.8	3	II	57.9	51.4	108.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1135	NM_001001481	Huesken	FLJ11011	55284	GAGAGCGCUGGGGACCAGUcu	ACUGGUCCCCAGCGCUCUC	-2.2	-2.4	-2.7	0	-1	4	-46.6	31.7	-4	1	III	68.4	37.4	93.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1136	NM_001001481	Huesken	FLJ11011	55284	GCGCUGGGGACCAGUCUUCug	GAAGACUGGUCCCCAGCGC	-2.4	-3.4	-2.7	1	-1	4	-45.4	28.4	0.1	2	II	68.4	36.9	80.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1137	NM_001001481	Huesken	FLJ11011	55284	CGCUGGGGACCAGUCUUCUgu	AGAAGACUGGUCCCCAGCG	-2.1	-2.4	-2.7	-1	-2	4	-44.1	37.8	-6	1	III	63.2	34.2	88.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1138	NM_001001481	Huesken	FLJ11011	55284	GGGGACCAGUCUUCUGUUAga	UAACAGAAGACUGGUCCCC	-1.3	-3.3	-1.5	1	-4	4	-40.6	46.7	-0.4	1	III	52.6	22.1	64.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1139	NM_001001481	Huesken	FLJ11011	55284	GGACCAGUCUUCUGUUAGAau	UCUAACAGAAGACUGGUCC	-2.4	-3.3	-1.5	1	-2	2	-38.5	47.8	-6.1	1	III	47.4	27.4	73.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1140	NM_001001481	Huesken	FLJ11011	55284	AGUCUUCUGUUAGAAUGGAua	UCCAUUCUAACAGAAGACU	-2.4	-2.1	0	3	1	2	-35	56.7	-3.8	4	II	36.8	42.4	64.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1141	NM_001001481	Huesken	FLJ11011	55284	UCUUCUGUUAGAAUGGAUAaa	UAUCCAUUCUAACAGAAGA	-1.3	-2.4	0	0	0	2	-33.1	85.1	-4	7	II	31.6	59.3	83.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1142	NM_001001481	Huesken	FLJ11011	55284	CUGUUAGAAUGGAUAAACAga	UGUUUAUCCAUUCUAACAG	-2.1	-2.1	0.2	-1	-2	2	-31.4	68.8	-6	4	II	31.6	45.1	66.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1143	NM_001001481	Huesken	FLJ11011	55284	UUAGAAUGGAUAAACAGAUau	AUCUGUUUAUCCAUUCUAA	-1.1	-0.9	2.1	2	2	2	-30.6	70.3	-3.9	8	II	26.3	70.4	96.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1144	NM_001001481	Huesken	FLJ11011	55284	UAGAAUGGAUAAACAGAUAug	UAUCUGUUUAUCCAUUCUA	-1.3	-1.3	2.1	0	-1	2	-31	73.5	-1.3	7	II	26.3	59.8	85.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1145	NM_001001481	Huesken	FLJ11011	55284	UGACCAUUGCUAUAAACAUga	AUGUUUAUAGCAAUGGUCA	-1.1	-2.1	-0.2	3	1	2	-32.6	70.8	-4	5	II	31.6	58.1	96.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1146	NM_001001481	Huesken	FLJ11011	55284	UUGCUAUAAACAUGAGGAUga	AUCCUCAUGUUUAUAGCAA	-1.1	-0.9	-0.2	1	1	2	-32.8	71.7	-6.3	7	II	31.6	57.7	104.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1147	NM_001001481	Huesken	FLJ11011	55284	GCUAUAAACAUGAGGAUGAac	UCAUCCUCAUGUUUAUAGC	-2.4	-3.4	1.6	0	-2	2	-34.3	59	-6.3	6	II	36.8	39.5	78.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1148	NM_001001481	Huesken	FLJ11011	55284	UAUAAACAUGAGGAUGAACag	GUUCAUCCUCAUGUUUAUA	-2.2	-1.3	1.6	0	3	2	-31.9	75.3	12.4	7	Ia	31.6	64.9	105.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1149	NM_001001481	Huesken	FLJ11011	55284	UAAACAUGAGGAUGAACAGga	CUGUUCAUCCUCAUGUUUA	-2.1	-1.3	1.6	0	4	2	-33.7	73.7	0	7	Ia	36.8	66	112.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1150	NM_001001481	Huesken	FLJ11011	55284	AACAUGAGGAUGAACAGGAau	UCCUGUUCAUCCUCAUGUU	-2.4	-0.9	1.6	2	1	2	-37.2	58.5	-6.3	7	II	42.1	58.7	108.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1151	NM_001001481	Huesken	FLJ11011	55284	ACAUGAGGAUGAACAGGAAua	UUCCUGUUCAUCCUCAUGU	-0.9	-2.2	1.6	1	0	2	-37.2	44.2	-6.3	5	II	42.1	34.3	73	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1152	NM_001001481	Huesken	FLJ11011	55284	AUGAGGAUGAACAGGAAUAuu	UAUUCCUGUUCAUCCUCAU	-1.3	-1.1	1.8	1	-1	2	-35.3	62.1	-6.4	6	II	36.8	55.3	104	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1153	NM_001001481	Huesken	FLJ11011	55284	GGAUGAACAGGAAUAUUUUca	AAAAUAUUCCUGUUCAUCC	-0.9	-3.3	2.1	0	-1	2	-31.4	64.8	-4.2	4	II	31.6	39.3	53.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1154	NM_001001481	Huesken	FLJ11011	55284	UGAACAGGAAUAUUUUCACca	GUGAAAAUAUUCCUGUUCA	-2.2	-2.1	1.1	0	4	2	-31.3	83.5	9.7	7	Ib	31.6	67.9	82	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1155	NM_001001481	Huesken	FLJ11011	55284	AACAGGAAUAUUUUCACCAgu	UGGUGAAAAUAUUCCUGUU	-2.1	-0.9	0.5	1	2	2	-32.2	80.9	1.6	6	II	31.6	65.8	115.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1156	NM_001001481	Huesken	FLJ11011	55284	ACAGGAAUAUUUUCACCAGua	CUGGUGAAAAUAUUCCUGU	-2.1	-2.2	0.5	1	3	2	-33.4	63.7	3	4	Ib	36.8	53.9	107.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1157	NM_001001481	Huesken	FLJ11011	55284	AGGAAUAUUUUCACCAGUAaa	UACUGGUGAAAAUAUUCCU	-1.3	-2.1	0.5	1	-1	2	-32.6	68	-6	5	II	31.6	53.1	94.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1158	NM_001001481	Huesken	FLJ11011	55284	GAAUAUUUUCACCAGUAAAca	UUUACUGGUGAAAAUAUUC	-0.9	-2.4	0.6	1	-2	2	-29	81.5	1.3	5	II	26.3	43.6	91.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1159	NM_001001481	Huesken	FLJ11011	55284	UUCACCAGUAAACAUGACCug	GGUCAUGUUUACUGGUGAA	-3.3	-0.9	-1.1	0	3	2	-35.8	82	12.1	6	Ib	42.1	78.4	103.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1160	NM_001001481	Huesken	FLJ11011	55284	GACCUGAGGAGAGUCAAAAgg	UUUUGACUCUCCUCAGGUC	-0.9	-2.4	-1	2	-2	2	-38.3	49.8	-1.4	3	III	47.4	35.1	95	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1161	NM_001001481	Huesken	FLJ11011	55284	CUGAGGAGAGUCAAAAGGAua	UCCUUUUGACUCUCCUCAG	-2.4	-2.1	0.2	-1	-2	2	-38.2	41.5	-10.7	3	III	47.4	45.5	78.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1162	NM_001001481	Huesken	FLJ11011	55284	GAGAGUCAAAAGGAUAUCGac	CGAUAUCCUUUUGACUCUC	-2.4	-2.4	1	0	1	2	-34.5	65.7	2.3	4	II	42.1	54.2	85.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1163	NM_001001481	Huesken	FLJ11011	55284	AGAGUCAAAAGGAUAUCGAcu	UCGAUAUCCUUUUGACUCU	-2.4	-2.1	-0.3	1	2	2	-34.5	66.5	-4	6	II	36.8	52.9	74.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1164	NM_001001481	Huesken	FLJ11011	55284	AGUCAAAAGGAUAUCGACUac	AGUCGAUAUCCUUUUGACU	-2.1	-2.1	-4.1	3	2	2	-34.3	67.5	3.8	4	II	36.8	56.1	81.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1165	NM_001001481	Huesken	FLJ11011	55284	CAAAAGGAUAUCGACUACUaa	AGUAGUCGAUAUCCUUUUG	-2.1	-2.1	1.9	-1	-1	2	-33.2	58.7	-5.6	6	II	36.8	57.6	123.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1166	NM_001001481	Huesken	FLJ11011	55284	UAUCGACUACUAAAUUUAAau	UUAAAUUUAGUAGUCGAUA	-0.9	-1.3	2.6	1	0	2	-27.7	76.3	-6.3	6	II	21.1	53.8	79.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1167	NM_001001481	Huesken	FLJ11011	55284	AUCGACUACUAAAUUUAAAua	UUUAAAUUUAGUAGUCGAU	-0.9	-1.1	2.9	3	0	2	-27.3	82.5	1	5	II	21.1	51.3	88	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1168	NM_001001481	Huesken	FLJ11011	55284	GACUACUAAAUUUAAAUAGaa	CUAUUUAAAUUUAGUAGUC	-2.1	-2.4	-0.9	1	0	1	-25.9	92.5	5.7	6	II	21.1	54.7	83.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1169	NM_001001481	Huesken	FLJ11011	55284	CUAAAUUUAAAUAGAAGUUga	AACUUCUAUUUAAAUUUAG	-0.9	-2.1	1	-1	-1	1	-24.2	89.4	-0.5	6	II	15.8	57.8	96.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1170	NM_001001481	Huesken	FLJ11011	55284	UUAAAUAGAAGUUGAAAUUuu	AAUUUCAACUUCUAUUUAA	-0.9	-0.9	4.1	-1	1	1	-25.3	96.5	1.8	9	II	15.8	72.6	97.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1171	NM_001001481	Huesken	FLJ11011	55284	UUCCCCUUCAUAUAAGGUAcc	UACCUUAUAUGAAGGGGAA	-1.3	-0.9	-3.1	3	0	4	-35.3	77.2	-3.8	5	II	36.8	62.4	111.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1172	NM_001001481	Huesken	FLJ11011	55284	AUAUAAGGUACCUGGUGCAcc	UGCACCAGGUACCUUAUAU	-2.1	-1.1	0.7	0	2	2	-37.4	60.7	-5.8	8	II	42.1	53.4	78	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1173	NM_001001481	Huesken	FLJ11011	55284	UAAGGUACCUGGUGCACCUuc	AGGUGCACCAGGUACCUUA	-2.1	-1.3	-6.5	0	3	2	-41.5	60.9	-4	5	II	52.6	66.3	93.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1174	NM_001001481	Huesken	FLJ11011	55284	UACCUGGUGCACCUUCCAUgu	AUGGAAGGUGCACCAGGUA	-1.1	-1.3	-1.5	2	2	2	-41.5	63.3	0.7	5	II	52.6	54.6	86.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1175	NM_001001481	Huesken	FLJ11011	55284	CUGGUGCACCUUCCAUGUCua	GACAUGGAAGGUGCACCAG	-2.4	-2.1	-1.5	0	0	2	-41.4	49.3	10.8	0	II	57.9	47	94.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1176	NM_001001481	Huesken	FLJ11011	55284	GGUGCACCUUCCAUGUCUAca	UAGACAUGGAAGGUGCACC	-1.3	-3.3	0.2	0	-1	2	-40.6	45.7	-8.7	1	III	52.6	24.8	38.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1177	NM_001001481	Huesken	FLJ11011	55284	CAUGUCUACAAUCCACUGUgu	ACAGUGGAUUGUAGACAUG	-2.2	-2.1	0	1	-1	2	-36	77.8	1.8	6	II	42.1	53	110.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1178	NM_001001481	Huesken	FLJ11011	55284	AUGUCUACAAUCCACUGUGua	CACAGUGGAUUGUAGACAU	-2.1	-1.1	-0.1	1	3	2	-36	76.6	0.1	6	Ia	42.1	68.4	115.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1179	NM_001001481	Huesken	FLJ11011	55284	UGUCUACAAUCCACUGUGUaa	ACACAGUGGAUUGUAGACA	-2.2	-2.1	-0.1	1	1	2	-37.1	64.6	-11.3	6	II	42.1	50.9	99.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1180	NM_001001481	Huesken	FLJ11011	55284	UCUACAAUCCACUGUGUAAuu	UUACACAGUGGAUUGUAGA	-0.9	-2.4	-1.1	0	-1	2	-35	76.6	-3.8	6	II	36.8	47.8	92.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1181	NM_001001481	Huesken	FLJ11011	55284	UACAAUCCACUGUGUAAUUga	AAUUACACAGUGGAUUGUA	-0.9	-1.3	-0.1	-1	0	2	-32.5	77.6	-1.3	6	II	31.6	57.8	111.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1182	NM_001001481	Huesken	FLJ11011	55284	CACUGUGUAAUUGAAUUUUga	AAAAUUCAAUUACACAGUG	-0.9	-2.1	2.8	-1	-3	1	-28.4	78.7	-3	5	III	26.3	48.9	76.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1183	NM_001001481	Huesken	FLJ11011	55284	GUGUAAUUGAAUUUUGAACac	GUUCAAAAUUCAAUUACAC	-2.2	-2.2	2.1	2	1	1	-27.5	95.7	17.8	5	II	26.3	55.2	73.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1184	NM_001001481	Huesken	FLJ11011	55284	AUUGAAUUUUGAACACUCUuc	AGAGUGUUCAAAAUUCAAU	-2.1	-1.1	-1.8	2	2	1	-29.7	78.7	-3.9	6	II	26.3	63.7	109.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1185	NM_001001481	Huesken	FLJ11011	55284	UUGAAUUUUGAACACUCUUcu	AAGAGUGUUCAAAAUUCAA	-0.9	-0.9	-1.8	1	1	1	-29.5	89.3	3.4	6	II	26.3	73.7	113.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1186	NM_001001481	Huesken	FLJ11011	55284	AAUUUUGAACACUCUUCUCau	GAGAAGAGUGUUCAAAAUU	-2.4	-0.9	0.2	1	4	1	-31	92.9	7.4	7	Ia	31.6	67.5	115.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1187	NM_001001481	Huesken	FLJ11011	55284	CUCUUCUCAUUUAAGGUCAuu	UGACCUUAAAUGAGAAGAG	-2.1	-2.1	0.3	0	-3	2	-33.7	76.9	-5.3	5	II	36.8	48.8	75	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1188	NM_001001481	Huesken	FLJ11011	55284	UCUUCUCAUUUAAGGUCAUuc	AUGACCUUAAAUGAGAAGA	-1.1	-2.4	0.3	-1	1	2	-32.7	70.6	-6.6	5	II	31.6	48.9	68.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1189	NM_001001481	Huesken	FLJ11011	55284	CUCAUUUAAGGUCAUUCCAgg	UGGAAUGACCUUAAAUGAG	-2.1	-2.1	0	0	-2	2	-33.6	79.1	-0.3	4	II	36.8	50.4	96.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1190	NM_001001481	Huesken	FLJ11011	55284	UUUAAGGUCAUUCCAGGAGgu	CUCCUGGAAUGACCUUAAA	-2.1	-0.9	0	1	4	2	-35.8	77.3	8.1	9	Ia	42.1	71.4	100.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1191	NM_001001481	Huesken	FLJ11011	55284	UUAAGGUCAUUCCAGGAGGug	CCUCCUGGAAUGACCUUAA	-3.3	-0.9	-0.8	1	4	2	-38.2	67.1	0.1	7	Ia	47.4	74.7	99.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1192	NM_016021	Huesken	UBE2J1	51465	AAUAUGUAUUCGUUUGCCAga	UGGCAAACGAAUACAUAUU	-2.1	-0.9	1.5	2	3	3	-31.7	85.3	-3.8	7	II	31.6	61	110	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1193	NM_016021	Huesken	UBE2J1	51465	UGUAUUCGUUUGCCAGAUAua	UAUCUGGCAAACGAAUACA	-1.3	-2.1	1.2	0	-1	3	-34.2	70.5	-1.4	6	II	36.8	50.6	91.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1194	NM_016021	Huesken	UBE2J1	51465	AUUCGUUUGCCAGAUAUAUuc	AUAUAUCUGGCAAACGAAU	-1.1	-1.1	1.2	4	1	3	-32.1	73	-6.3	6	II	31.6	56.1	95.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1195	NM_016021	Huesken	UBE2J1	51465	AUUCGUCGGAAUAUAAGAGcu	CUCUUAUAUUCCGACGAAU	-2.1	-1.1	0.6	3	5	3	-32.7	74.9	5.1	5	Ib	36.8	60.5	98.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1196	NM_016021	Huesken	UBE2J1	51465	CUGCCAAUGCCAAAGUCAGga	CUGACUUUGGCAUUGGCAG	-2.1	-2.1	-0.9	1	0	3	-38.6	46.8	-6.9	2	II	52.6	45.7	76.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1197	NM_016021	Huesken	UBE2J1	51465	GGAUGACAAUCAGUACAGCug	GCUGUACUGAUUGUCAUCC	-3.4	-3.3	1.2	0	2	2	-37.5	45.8	7.4	2	II	47.4	37.5	62.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1198	NM_016021	Huesken	UBE2J1	51465	CAGUACAGCUGACCCACCAug	UGGUGGGUCAGCUGUACUG	-2.1	-2.1	-0.4	0	-2	3	-42.6	48.8	-0.7	3	II	57.9	42.4	83.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1199	NM_016021	Huesken	UBE2J1	51465	GGUUGUCUCUUGGCUGGUGgc	CACCAGCCAAGAGACAACC	-2.1	-3.3		1	1	3	-41.2	49	0	1	II	57.9	35	55.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1200	NM_016021	Huesken	UBE2J1	51465	UUGGCUGGUGGCCCUGGAUua	AUCCAGGGCCACCAGCCAA	-1.1	-0.9	-3.8	0	1	5	-45.8	51.8	-3.9	3	II	63.2	52.6	76	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1201	NM_016021	Huesken	UBE2J1	51465	ACAUCUGGUGAAGUAGACAac	UGUCUACUUCACCAGAUGU	-2.1	-2.2	-1	1	1	2	-37.3	60	0.9	4	II	42.1	50	73.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1202	NM_016021	Huesken	UBE2J1	51465	GAAGUAGACAACCUUCUCUga	AGAGAAGGUUGUCUACUUC	-2.1	-2.4	-2.4	1	1	2	-36.1	68.7	0.7	5	II	42.1	52.1	78.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1203	NM_016021	Huesken	UBE2J1	51465	ACUCUGCUGCUGGGCCCGGcg	CCGGGCCCAGCAGCAGAGU	-3.3	-2.2	-0.7	2	4	8	-48.4	26.1	0	3	II	73.7	44.4	47.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1204	NM_016021	Huesken	UBE2J1	51465	CUGCUGCUGGGCCCGGCGCug	GCGCCGGGCCCAGCAGCAG	-3.4	-2.1	-0.8	1	1	11	-50.9	29.8	5.5	1	II	84.2	43.6	35.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1205	NM_016021	Huesken	UBE2J1	51465	UUCUUAGCUACAGGUUGGGua	CCCAACCUGUAGCUAAGAA	-3.3	-0.9	1	-1	4	3	-38	63.6	-2.4	8	Ia	47.4	63.9	78.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1206	NM_016021	Huesken	UBE2J1	51465	AGGUUGGGUAGGUUGAUGAaa	UCAUCAACCUACCCAACCU	-2.4	-2.1		1	0	3	-39.3	67.9	-1.4	7	II	47.4	49.4	47.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1207	NM_016021	Huesken	UBE2J1	51465	UGGGUAGGUUGAUGAAAGGau	CCUUUCAUCAACCUACCCA	-3.3	-2.1	4.4	-1	2	3	-38	54.1	0	5	II	47.4	56.6	56.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1208	NM_016021	Huesken	UBE2J1	51465	UGGCCGUAGCACCCUGGAAug	UUCCAGGGUGCUACGGCCA	-0.9	-2.1	-2.9	3	-1	5	-45.2	45.6	-1.4	2	II	63.2	41	75.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1209	NM_016021	Huesken	UBE2J1	51465	GGUAUAUCAUCUUGUAAAUca	AUUUACAAGAUGAUAUACC	-1.1	-3.3	2.2	0	-2	2	-29.8	79.5	-3.3	4	II	26.3	33.6	47.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1210	NM_016021	Huesken	UBE2J1	51465	UUGUAAAUCAGUUAGUGAAaa	UUCACUAACUGAUUUACAA	-0.9	-0.9	0.6	1	0	1	-30.1	91.1	2	8	II	26.3	61.4	67.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1211	NM_016021	Huesken	UBE2J1	51465	AGUGGUUUAAGUCUGACUCag	GAGUCAGACUUAAACCACU	-2.4	-2.1	0.1	2	4	2	-35.9	83.7	15.2	5	Ib	42.1	65.3	88.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1212	NM_016021	Huesken	UBE2J1	51465	UUCCAGAUGAAUUGACUUCug	GAAGUCAAUUCAUCUGGAA	-2.4	-0.9	-0.5	1	2	2	-33.8	80.2	15.4	8	Ib	36.8	71.6	75.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1213	NM_016021	Huesken	UBE2J1	51465	UCCAGAUGAAUUGACUUCUgc	AGAAGUCAAUUCAUCUGGA	-2.1	-2.4	-0.7	1	0	2	-35	72.8	-1	7	II	36.8	63.2	79.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1214	NM_016021	Huesken	UBE2J1	51465	CCAGAUGAAUUGACUUCUGcc	CAGAAGUCAAUUCAUCUGG	-2.1	-3.3	-1.7	-1	0	2	-34.7	52.1	-1.9	2	II	42.1	47.2	60.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1215	NM_016021	Huesken	UBE2J1	51465	CAGAUGAAUUGACUUCUGCcu	GCAGAAGUCAAUUCAUCUG	-3.4	-2.1	-0.7	-1	1	2	-34.8	71.7	10.1	5	II	42.1	57.5	54.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1216	NM_016021	Huesken	UBE2J1	51465	CUGCCUUAAAGCUUAUUUGcc	CAAAUAAGCUUUAAGGCAG	-2.1	-2.1	-2.4	1	0	3	-31.8	85.3	-2	3	II	36.8	58.1	56	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1217	NM_016021	Huesken	UBE2J1	51465	GUUCUUUGGCUUCUUGGUCag	GACCAAGAAGCCAAAGAAC	-2.4	-2.2		3	3	3	-36.6	72.8	15.2	4	II	47.4	53.3	56.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1218	NM_016021	Huesken	UBE2J1	51465	UCUUUGGCUUCUUGGUCAGcu	CUGACCAAGAAGCCAAAGA	-2.1	-2.4	1	-1	3	3	-37.7	70	0.7	7	Ib	47.4	55	73.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1219	NM_016021	Huesken	UBE2J1	51465	GCUUCCAGAUUUUAAAGGCaa	GCCUUUAAAAUCUGGAAGC	-3.4	-3.4	-0.1	0	2	3	-34.4	69.2	10.4	3	II	42.1	46	50.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1220	NM_016021	Huesken	UBE2J1	51465	UUCCAGAUUUUAAAGGCAAca	UUGCCUUUAAAAUCUGGAA	-0.9	-0.9	-0.1	1	0	3	-31.9	63.9	-3.6	6	II	31.6	58.3	70.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1221	NM_016021	Huesken	UBE2J1	51465	UCCAGAUUUUAAAGGCAACag	GUUGCCUUUAAAAUCUGGA	-2.2	-2.4	-0.1	0	1	3	-33.2	67.4	7.1	5	Ib	36.8	57.6	55.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1222	NM_016021	Huesken	UBE2J1	51465	GCAAGUGCUCUUCUUUCCUca	AGGAAAGAAGAGCACUUGC	-2.1	-3.4	0.2	0	1	2	-37.8	61	3.8	3	III	47.4	44.1	70.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1223	NM_016021	Huesken	UBE2J1	51465	AACCUAUGGCUCCCUCUCCuu	GGAGAGGGAGCCAUAGGUU	-3.3	-0.9	-0.7	4	3	3	-43.1	59	7.5	5	Ia	57.9	62.3	61	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1224	NM_016021	Huesken	UBE2J1	51465	UGGCUCCCUCUCCUUUUGUug	ACAAAAGGAGAGGGAGCCA	-2.2	-2.1	-0.9	1	1	3	-41.3	64.9	-1.6	5	II	52.6	52.3	41.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1225	NM_016021	Huesken	UBE2J1	51465	CCCUCUCCUUUUGUUGGCAua	UGCCAACAAAAGGAGAGGG	-2.1	-3.3	-0.2	-1	-2	3	-39.8	55.7	-0.4	1	III	52.6	30.2	28.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1226	NM_016021	Huesken	UBE2J1	51465	UGAUGGCUAAUAAUGCUGUcc	ACAGCAUUAUUAGCCAUCA	-2.2	-2.1	0	0	2	3	-35	72.3	-4.2	7	II	36.8	63.1	54.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1227	NM_016021	Huesken	UBE2J1	51465	UGGCUAAUAAUGCUGUCCUua	AGGACAGCAUUAUUAGCCA	-2.1	-2.1	0	1	2	3	-37.2	73.4	0.8	6	II	42.1	63.8	74.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1228	NM_016021	Huesken	UBE2J1	51465	UAAUGCUGUCCUUAUACUCca	GAGUAUAAGGACAGCAUUA	-2.4	-1.3	0.7	1	4	2	-34.3	87	8.1	7	Ia	36.8	76.1	91.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1229	NM_016021	Huesken	UBE2J1	51465	AGGAUGAUGGCCUGAGAUGcu	CAUCUCAGGCCAUCAUCCU	-2.1	-2.1	0.2	1	1	4	-40.9	52.6	-4.7	5	Ib	52.6	52.7	64.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1230	NM_016021	Huesken	UBE2J1	51465	AUGAUGGCCUGAGAUGCUCaa	GAGCAUCUCAGGCCAUCAU	-2.4	-1.1	-0.5	2	3	4	-41	55	7.4	6	Ib	52.6	61.4	64.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1231	NM_016021	Huesken	UBE2J1	51465	UGGCCUGAGAUGCUCAAACag	GUUUGAGCAUCUCAGGCCA	-2.2	-2.1	-5.3	2	2	4	-40.4	64	12.1	4	II	52.6	63.5	72.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1232	NM_016021	Huesken	UBE2J1	51465	CCUGAGAUGCUCAAACAGAuu	UCUGUUUGAGCAUCUCAGG	-2.4	-3.3	-2	0	-3	2	-38.2	39.2	-8	2	III	47.4	38.8	56	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1233	NM_016021	Huesken	UBE2J1	51465	CCCACUUCAAAUCGACCAUua	AUGGUCGAUUUGAAGUGGG	-1.1	-3.3	1.6	-1	-3	3	-37.2	62.4	1.8	3	III	47.4	34.6	56.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1234	NM_016021	Huesken	UBE2J1	51465	CACUUCAAAUCGACCAUUAgc	UAAUGGUCGAUUUGAAGUG	-1.3	-2.1	1.6	-1	-4	2	-32.8	68.4	-10.7	5	II	36.8	41.4	46.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1235	NM_016021	Huesken	UBE2J1	51465	AAUCGACCAUUAGCCGUUAgg	UAACGGCUAAUGGUCGAUU	-1.3	-0.9	-4.2	2	0	4	-35.7	51.1	-3.8	4	II	42.1	42.4	84	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1236	NM_016021	Huesken	UBE2J1	51465	UAAUGCUUGGUGGUUUCAUgg	AUGAAACCACCAAGCAUUA	-1.1	-1.3	1.7	2	3	2	-34.4	78	-1.6	7	II	36.8	62.4	68.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1237	NM_016021	Huesken	UBE2J1	51465	CAUGGGAUACUCUGGUGGCag	GCCACCAGAGUAUCCCAUG	-3.4	-2.1	1.9	-2	2	3	-41.9	48.5	3.1	2	II	57.9	49.6	53.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1238	NM_016021	Huesken	UBE2J1	51465	ACUAUCCGCCCGUGAUAAAcu	UUUAUCACGGGCGGAUAGU	-0.9	-2.2	-1.2	1	-1	7	-38.1	54.2	-6.1	4	II	47.4	31.6	51.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1239	NM_016021	Huesken	UBE2J1	51465	ACUCCUCCAUCAAAAUCGGag	CCGAUUUUGAUGGAGGAGU	-3.3	-2.2	0.1	2	4	3	-37.4	54.8	-4.9	3	II	47.4	51.4	48	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1240	NM_016021	Huesken	UBE2J1	51465	GGCUGCGCAUGGUAAUGAUcu	AUCAUUACCAUGCGCAGCC	-1.1	-3.3	1.5	0	-2	4	-39.8	48	-4	2	III	52.6	21.7	55.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1241	NM_005339	Huesken	HIP2	3093	GAGCUGUCUGUUUGAACAUuu	AUGUUCAAACAGACAGCUC	-1.1	-2.4	-1	1	-1	2	-35.7	59	-1	2	III	42.1	40	64.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1242	NM_005339	Huesken	HIP2	3093	CUGUCUGUUUGAACAUUUCgg	GAAAUGUUCAAACAGACAG	-2.4	-2.1	-1	-1	-1	1	-32	76.5	2.8	3	II	36.8	54.6	76.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1243	NM_005339	Huesken	HIP2	3093	UCUGUUUGAACAUUUCGGGau	CCCGAAAUGUUCAAACAGA	-3.3	-2.4	0.4	1	5	4	-34.6	81.9	0	7	Ia	42.1	71.6	92.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1244	NM_005339	Huesken	HIP2	3093	UUUGAACAUUUCGGGAUUUug	AAAUCCCGAAAUGUUCAAA	-0.9	-0.9	0.9	-1	1	4	-31.1	64.6	-3.6	7	II	31.6	56.5	73.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1245	NM_005339	Huesken	HIP2	3093	UUUCGGGAUUUUGUUUGUAcu	UACAAACAAAAUCCCGAAA	-1.3	-0.9		1	2	4	-31.1	71.6	1.6	6	II	31.6	57.2	66.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1246	NM_005339	Huesken	HIP2	3093	UUCGGGAUUUUGUUUGUACug	GUACAAACAAAAUCCCGAA	-2.2	-0.9	1.2	0	3	4	-32.4	81.7	9.7	6	II	36.8	69.2	73.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1247	NM_005339	Huesken	HIP2	3093	CGGGAUUUUGUUUGUACUGau	CAGUACAAACAAAAUCCCG	-2.1	-2.4	1	-1	-1	4	-33.3	65.6	3.7	2	II	42.1	50.1	70.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1248	NM_005339	Huesken	HIP2	3093	UUGUUUGUACUGAUUUGCUac	AGCAAAUCAGUACAAACAA	-2.1	-0.9	0.9	0	3	2	-31.9	89.4	-4.2	7	II	31.6	71.4	77.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1249	NM_005339	Huesken	HIP2	3093	UUGUACUGAUUUGCUACUAca	UAGUAGCAAAUCAGUACAA	-1.3	-0.9	0.8	1	0	2	-32.7	79.5	2	7	II	31.6	60.1	87.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1250	NM_005339	Huesken	HIP2	3093	UGUACUGAUUUGCUACUACag	GUAGUAGCAAAUCAGUACA	-2.2	-2.1	0.8	0	3	2	-34	82.2	12.1	7	Ib	36.8	64.6	92.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1251	NM_005339	Huesken	HIP2	3093	ACUGAUUUGCUACUACAGCau	GCUGUAGUAGCAAAUCAGU	-3.4	-2.2	-2	3	4	2	-36	68.1	14.8	4	Ia	42.1	63.3	87.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1252	NM_005339	Huesken	HIP2	3093	GAUUUGCUACUACAGCAUCcu	GAUGCUGUAGUAGCAAAUC	-2.4	-2.4	-3	0	1	2	-35.2	66	9.8	3	II	42.1	49.7	65.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1253	NM_005339	Huesken	HIP2	3093	UUUGCUACUACAGCAUCCUgu	AGGAUGCUGUAGUAGCAAA	-2.1	-0.9	-3	0	3	2	-37.1	72.7	-6.3	6	II	42.1	71.9	92.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1254	NM_005339	Huesken	HIP2	3093	UUGCUACUACAGCAUCCUGug	CAGGAUGCUGUAGUAGCAA	-2.1	-0.9	-3	1	3	2	-38.3	66.2	4.7	4	Ib	47.4	63.1	93	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1255	NM_005339	Huesken	HIP2	3093	UGCUACUACAGCAUCCUGUgg	ACAGGAUGCUGUAGUAGCA	-2.2	-2.1	-2.4	2	1	2	-39.6	76.3	-3.9	7	II	47.4	62.8	89.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1256	NM_005339	Huesken	HIP2	3093	CUACAGCAUCCUGUGGAUCau	GAUCCACAGGAUGCUGUAG	-2.4	-2.1	-2.7	0	1	2	-39.8	50.5	11.1	2	II	52.6	48.5	67.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1257	NM_005339	Huesken	HIP2	3093	AGCAUCCUGUGGAUCAUCUgg	AGAUGAUCCACAGGAUGCU	-2.1	-2.1	-1.7	2	1	2	-39.8	68.4	-4.2	4	II	47.4	49.3	68	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1258	NM_005339	Huesken	HIP2	3093	CAUCCUGUGGAUCAUCUGGcu	CCAGAUGAUCCACAGGAUG	-3.3	-2.1	-1.7	1	1	2	-39.7	67.1	5.7	3	II	52.6	53.5	66.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1259	NM_005339	Huesken	HIP2	3093	AUCCUGUGGAUCAUCUGGCuc	GCCAGAUGAUCCACAGGAU	-3.4	-1.1	-1	4	5	3	-41	63.6	4.8	5	Ib	52.6	68.5	74.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1260	NM_005339	Huesken	HIP2	3093	UGUGGAUCAUCUGGCUCUGca	CAGAGCCAGAUGAUCCACA	-2.1	-2.1	1.3	0	3	3	-40.8	54.3	0.7	5	Ib	52.6	55	82.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1261	NM_005339	Huesken	HIP2	3093	AUCUGGCUCUGCAGCUGCCaa	GGCAGCUGCAGAGCCAGAU	-3.3	-1.1	-3.5	1	4	3	-45	49	2.5	2	II	63.2	54.3	81.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1262	NM_005339	Huesken	HIP2	3093	UGGCUCUGCAGCUGCCAAUag	AUUGGCAGCUGCAGAGCCA	-1.1	-2.1	-3.6	2	-1	3	-43.5	58.2	-6.3	4	II	57.9	50.7	85	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1263	NM_005339	Huesken	HIP2	3093	CUCUGCAGCUGCCAAUAGUgc	ACUAUUGGCAGCUGCAGAG	-2.2	-2.1	-1.3	0	-2	3	-40.3	53.8	-5.9	2	III	52.6	40.1	62	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1264	NM_005339	Huesken	HIP2	3093	UAGUGCUUGCAAUGACAAUaa	AUUGUCAUUGCAAGCACUA	-1.1	-1.3	-0.5	1	0	2	-34.5	73.6	-1.7	6	II	36.8	55.4	85.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1265	NM_005339	Huesken	HIP2	3093	AGUGCUUGCAAUGACAAUAau	UAUUGUCAUUGCAAGCACU	-1.3	-2.1	-0.5	3	-1	2	-34.5	74.2	1.6	5	II	36.8	54	89.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1266	NM_005339	Huesken	HIP2	3093	UGACAAUAAUACCGUGCGGag	CCGCACGGUAUUAUUGUCA	-3.3	-2.1	-0.3	1	4	4	-36.7	58.2	5.1	6	Ia	47.4	58.9	63.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1267	NM_005339	Huesken	HIP2	3093	AUACCGUGCGGAGAGUCAUug	AUGACUCUCCGCACGGUAU	-1.1	-1.1	-0.4	4	2	4	-40.5	50	-4	5	II	52.6	51.5	85.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1268	NM_005339	Huesken	HIP2	3093	UGCGGAGAGUCAUUGCAGCug	GCUGCAAUGACUCUCCGCA	-3.4	-2.1	0	1	3	4	-42	49.5	7.4	3	II	57.9	57.1	87.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1269	NM_005339	Huesken	HIP2	3093	GGAGAGUCAUUGCAGCUGCcc	GCAGCUGCAAUGACUCUCC	-3.4	-3.3	-1.3	1	1	2	-41.7	44.9	12.5	2	II	57.9	46.8	47.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1270	NM_005339	Huesken	HIP2	3093	GAGAGUCAUUGCAGCUGCCca	GGCAGCUGCAAUGACUCUC	-3.3	-2.4	-1.3	-1	2	3	-41.7	46.1	2.5	0	II	57.9	45	57.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1271	NM_005339	Huesken	HIP2	3093	GAGUCAUUGCAGCUGCCCAuu	UGGGCAGCUGCAAUGACUC	-2.1	-2.4	-1.3	2	0	4	-42.6	57.6	3.3	1	II	57.9	38.7	44.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1272	NM_005339	Huesken	HIP2	3093	AUUGCAGCUGCCCAUUGAUcu	AUCAAUGGGCAGCUGCAAU	-1.1	-1.1	-1.3	1	2	4	-39	57.3	-6.3	4	II	47.4	47.7	48.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1273	NM_005339	Huesken	HIP2	3093	GCUGCCCAUUGAUCUUUCAgg	UGAAAGAUCAAUGGGCAGC	-2.1	-3.4	0.5	0	-2	4	-38.1	58	-3.8	2	III	47.4	28.5	57.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1274	NM_005339	Huesken	HIP2	3093	CUGCCCAUUGAUCUUUCAGga	CUGAAAGAUCAAUGGGCAG	-2.1	-2.1	0.4	0	1	4	-36.8	73.3	8.1	2	II	47.4	49.4	79.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1275	NM_005339	Huesken	HIP2	3093	UCUUUCAGGAUAUCCAAACaa	GUUUGGAUAUCCUGAAAGA	-2.2	-2.4	-1	1	2	2	-33.9	82.9	7.4	7	Ia	36.8	69.9	87.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1276	NM_005339	Huesken	HIP2	3093	CUUUCAGGAUAUCCAAACAaa	UGUUUGGAUAUCCUGAAAG	-2.1	-2.1	-1	-1	-2	2	-33.6	71.6	2	3	II	36.8	42.4	58.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1277	NM_005339	Huesken	HIP2	3093	UUCAGGAUAUCCAAACAAAua	UUUGUUUGGAUAUCCUGAA	-0.9	-0.9	-1	0	-1	2	-32.4	67.3	-11	5	II	31.6	56.7	78.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1278	NM_005339	Huesken	HIP2	3093	UAUCCAAACAAAUAGCCCCug	GGGGCUAUUUGUUUGGAUA	-3.3	-1.3	-0.5	2	5	5	-35.9	79.3	9.7	7	Ia	42.1	80.5	81.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1279	NM_005339	Huesken	HIP2	3093	AUCCAAACAAAUAGCCCCUgu	AGGGGCUAUUUGUUUGGAU	-2.1	-1.1	-0.5	1	2	5	-36.7	61.8	1.5	6	II	42.1	58.6	71	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1280	NM_005339	Huesken	HIP2	3093	AAUAGCCCCUGUGACGGAAcu	UUCCGUCACAGGGGCUAUU	-0.9	-0.9	-2.1	2	1	5	-40.8	53.5	-0.7	4	II	52.6	39.4	55.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1281	NM_005339	Huesken	HIP2	3093	UGACGGAACUAAUAUUAGGau	CCUAAUAUUAGUUCCGUCA	-3.3	-2.1	1	2	3	3	-33.3	69.6	2.3	5	II	36.8	67.6	74.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1282	NM_005339	Huesken	HIP2	3093	ACGGAACUAAUAUUAGGAUgc	AUCCUAAUAUUAGUUCCGU	-1.1	-2.2	1	1	1	3	-32.3	63.4	1.1	4	II	31.6	43.7	53.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1283	NM_005339	Huesken	HIP2	3093	AACUAAUAUUAGGAUGCCAua	UGGCAUCCUAAUAUUAGUU	-2.1	-0.9	1	2	1	3	-32.9	69.9	1.3	6	II	31.6	52.7	61.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1284	NM_005339	Huesken	HIP2	3093	UAUUAGGAUGCCAUAUUUUag	AAAAUAUGGCAUCCUAAUA	-0.9	-1.3	1.2	0	2	3	-30.6	83.4	-6.2	9	II	26.3	67	67.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1285	NM_005339	Huesken	HIP2	3093	UAGGAUGCCAUAUUUUAGUga	ACUAAAAUAUGGCAUCCUA	-2.2	-1.3	1.2	1	2	3	-32.9	84.5	1.1	7	II	31.6	71.8	76	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1286	NM_005339	Huesken	HIP2	3093	CAUAUUUUAGUGAUAAACCgg	GGUUUAUCACUAAAAUAUG	-3.3	-2.1	1	0	2	2	-28.1	93.3	5.1	5	II	26.3	71.4	81.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1287	NM_005339	Huesken	HIP2	3093	UAUUUUAGUGAUAAACCGGac	CCGGUUUAUCACUAAAAUA	-3.3	-1.3	0.7	0	5	4	-30.6	90	2.7	8	Ia	31.6	80.7	89.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1288	NM_005339	Huesken	HIP2	3093	AUUUUAGUGAUAAACCGGAcc	UCCGGUUUAUCACUAAAAU	-2.4	-1.1	0.7	2	2	4	-31.7	70.3	-6.3	8	II	31.6	60.7	73	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1289	NM_005339	Huesken	HIP2	3093	AUUAAAUGGGUAUGUUUCUgg	AGAAACAUACCCAUUUAAU	-2.1	-1.1	3.1	2	2	3	-30.1	79.4	-1.3	8	II	26.3	62	72.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1290	NM_005339	Huesken	HIP2	3093	UAAAUGGGUAUGUUUCUGGua	CCAGAAACAUACCCAUUUA	-3.3	-1.3	3.1	0	5	3	-33.5	86.6	2.3	10	Ia	36.8	81.1	72.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1291	NM_005339	Huesken	HIP2	3093	AUGGGUAUGUUUCUGGUAUuu	AUACCAGAAACAUACCCAU	-1.1	-1.1	4.4	2	1	3	-35	75.1	3.8	5	II	36.8	54.8	76.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1292	NM_005339	Huesken	HIP2	3093	UUUUAUCUCUAGUUGGUAUcu	AUACCAACUAGAGAUAAAA	-1.1	-0.9	2.8	1	2	2	-30.6	101.7	3.4	8	II	26.3	62.2	78.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1293	NM_005339	Huesken	HIP2	3093	UCUAGUUGGUAUCUUCCUCcu	GAGGAAGAUACCAACUAGA	-2.4	-2.4	-0.9	2	4	2	-36.6	83.3	17.5	6	Ia	42.1	67.7	79.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1294	NM_005339	Huesken	HIP2	3093	UAGUUGGUAUCUUCCUCCUuc	AGGAGGAAGAUACCAACUA	-2.1	-1.3	-0.9	0	2	2	-37.5	82	-3.3	7	II	42.1	66.4	75.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1295	NM_005339	Huesken	HIP2	3093	AGUUGGUAUCUUCCUCCUUca	AAGGAGGAAGAUACCAACU	-0.9	-2.1	-0.9	2	1	2	-37.1	67.6	1.5	5	II	42.1	51.2	68.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1296	NM_005339	Huesken	HIP2	3093	UUGGUAUCUUCCUCCUUCAua	UGAAGGAGGAAGAUACCAA	-2.1	-0.9	-0.9	1	0	2	-37.3	76.8	-8.5	7	II	42.1	64.7	82.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1297	NM_005339	Huesken	HIP2	3093	CUUCCUCCUUCAUAUGGUGug	CACCAUAUGAAGGAGGAAG	-2.1	-2.1	-0.5	0	1	2	-37.2	62.4	-4.4	2	II	47.4	47.1	69.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1298	NM_005339	Huesken	HIP2	3093	CUCCUUCAUAUGGUGUGUCug	GACACACCAUAUGAAGGAG	-2.4	-2.1	-0.5	0	1	2	-37.4	63.1	7.7	3	II	47.4	51.3	64.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1299	NM_005339	Huesken	HIP2	3093	UCCUUCAUAUGGUGUGUCUgg	AGACACACCAUAUGAAGGA	-2.1	-2.4	-0.5	-1	0	2	-37.4	78.1	-4	7	II	42.1	53.3	81	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1300	NM_005339	Huesken	HIP2	3093	CCUUCAUAUGGUGUGUCUGga	CAGACACACCAUAUGAAGG	-2.1	-3.3	0.6	0	1	2	-37.1	68	3.4	3	II	47.4	44.8	72.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1301	NM_005339	Huesken	HIP2	3093	UUCAUAUGGUGUGUCUGGAgg	UCCAGACACACCAUAUGAA	-2.4	-0.9	1.8	2	2	2	-37.4	74.8	1.6	7	II	42.1	56.9	73.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1302	NM_005339	Huesken	HIP2	3093	CAUAUGGUGUGUCUGGAGGuc	CCUCCAGACACACCAUAUG	-3.3	-2.1	2.9	0	2	2	-39.5	68.4	5.8	5	II	52.6	51.9	66.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1303	NM_005339	Huesken	HIP2	3093	UGGUGUGUCUGGAGGUCCUgc	AGGACCUCCAGACACACCA	-2.1	-2.1	-1.7	0	1	2	-43.9	53.7	-6.6	3	II	57.9	52.9	74.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1304	NM_005339	Huesken	HIP2	3093	UGUCUGGAGGUCCUGCUAUuu	AUAGCAGGACCUCCAGACA	-1.1	-2.1	-1.7	2	1	2	-42	57.5	-1.6	6	II	52.6	52.9	69.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1305	NM_005339	Huesken	HIP2	3093	GUCUGGAGGUCCUGCUAUUuc	AAUAGCAGGACCUCCAGAC	-0.9	-2.2	-1.7	0	-2	2	-40.8	46.3	-8.7	2	III	52.6	33.2	56.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1306	NM_005339	Huesken	HIP2	3093	UCUGGAGGUCCUGCUAUUUcu	AAAUAGCAGGACCUCCAGA	-0.9	-2.4	-1.7	0	0	2	-39.5	55.3	-3.3	5	II	47.4	50.6	65.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1307	NM_005339	Huesken	HIP2	3093	AGGUCCUGCUAUUUCUCCUcu	AGGAGAAAUAGCAGGACCU	-2.1	-2.1	0.8	3	2	2	-39.5	80.7	3.7	4	II	47.4	53.8	69.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1308	NM_005339	Huesken	HIP2	3093	GGUCCUGCUAUUUCUCCUCuu	GAGGAGAAAUAGCAGGACC	-2.4	-3.3	2.2	1	1	2	-39.8	65.4	10.4	3	II	52.6	48.7	54.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1309	NM_005339	Huesken	HIP2	3093	CUAUUUCUCCUCUUAAUUCug	GAAUUAAGAGGAGAAAUAG	-2.4	-2.1	3.1	0	1	2	-30.6	90.6	5.4	5	II	31.6	62.5	74.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1310	NM_005339	Huesken	HIP2	3093	UCCUCUUAAUUCUGUAAAAuu	UUUUACAGAAUUAAGAGGA	-0.9	-2.4	0.8	0	-2	2	-30.2	83	-6.1	5	II	26.3	48.2	47.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1311	NM_005339	Huesken	HIP2	3093	UGUAAAAUUCUCAUCUACAag	UGUAGAUGAGAAUUUUACA	-2.1	-2.1	-2.3	0	1	1	-30.5	82.5	-3.7	6	II	26.3	65.6	89.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1312	NM_005339	Huesken	HIP2	3093	GUAAAAUUCUCAUCUACAAga	UUGUAGAUGAGAAUUUUAC	-0.9	-2.2	-0.6	2	-1	1	-29.3	75.6	-3.4	5	II	26.3	43.1	55.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1313	NM_005339	Huesken	HIP2	3093	UAAAAUUCUCAUCUACAAGau	CUUGUAGAUGAGAAUUUUA	-2.1	-1.3	0.2	1	4	1	-29.2	98.2	12.8	8	Ia	26.3	76.4	91.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1314	NM_005339	Huesken	HIP2	3093	AAAAUUCUCAUCUACAAGAuc	UCUUGUAGAUGAGAAUUUU	-2.4	-0.9	-1.4	2	1	1	-30.3	87.4	-6.1	6	II	26.3	66.5	84.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1315	NM_005339	Huesken	HIP2	3093	AAAUUCUCAUCUACAAGAUcu	AUCUUGUAGAUGAGAAUUU	-1.1	-0.9	-1.7	2	2	1	-30.5	85.6	-5.9	6	II	26.3	56.4	81.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1316	NM_005339	Huesken	HIP2	3093	UUCUCAUCUACAAGAUCUAcu	UAGAUCUUGUAGAUGAGAA	-1.3	-0.9	-1.7	0	0	1	-33.4	74.3	-8.7	6	II	31.6	58.1	85.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1317	NM_005339	Huesken	HIP2	3093	CUCAUCUACAAGAUCUACUuu	AGUAGAUCUUGUAGAUGAG	-2.1	-2.1	-0.2	1	-1	1	-34.4	83.5	-3.9	5	II	36.8	58.6	88.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1318	NM_005339	Huesken	HIP2	3093	UCAUCUACAAGAUCUACUUua	AAGUAGAUCUUGUAGAUGA	-0.9	-2.4	-0.2	0	1	1	-33.2	87.7	-4	7	II	31.6	65.7	88.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1319	NM_005339	Huesken	HIP2	3093	CUACAAGAUCUACUUUAAUuu	AUUAAAGUAGAUCUUGUAG	-1.1	-2.1	0.6	0	-1	1	-29.4	66.1	1.5	4	II	26.3	44.3	45.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1320	NM_053656	Huesken	P2rx2	22953	GGUCUUUGUCCCAUAUGCUgg	AGCAUAUGGGACAAAGACC	-2.1	-3.3	-0.1	2	1	3	-38.3	60.9	-8.9	3	II	47.4	42.5	53.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1321	NM_053656	Huesken	P2rx2	22953	GUCUUUGUCCCAUAUGCUGgc	CAGCAUAUGGGACAAAGAC	-2.1	-2.2	1.1	1	1	3	-37.1	68.3	-2.4	4	II	47.4	50.7	51.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1322	NM_053656	Huesken	P2rx2	22953	GUCCCAUAUGCUGGCCAAGug	CUUGGCCAGCAUAUGGGAC	-2.1	-2.2	-2.4	1	0	4	-41.6	43.4	0.7	0	II	57.9	36.7	52	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1323	NM_053656	Huesken	P2rx2	22953	GCCAAGUGUGUCCACCACCug	GGUGGUGGACACACUUGGC	-3.3	-3.4	-3	1	0	3	-43.5	41.2	10.2	1	II	63.2	49.1	52.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1324	NM_053656	Huesken	P2rx2	22953	ACCACCUGCUCAGUCAGAGca	CUCUGACUGAGCAGGUGGU	-2.1	-2.2	-1.8	3	3	2	-42.7	53.7	0	2	II	57.9	47.3	46.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1325	NM_053656	Huesken	P2rx2	22953	CUCAGUCAGAGCAGAGAUGga	CAUCUCUGCUCUGACUGAG	-2.1	-2.1	-1.1	-1	-1	2	-39.7	47.9	-6.9	2	II	52.6	44.8	42.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1326	NM_053656	Huesken	P2rx2	22953	CGAGGAGACUGGACAGCCAgu	UGGCUGUCCAGUCUCCUCG	-2.1	-2.4	-2.9	0	-2	3	-44.4	25.5	-8.3	0	III	63.2	32.1	67.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1327	NM_053656	Huesken	P2rx2	22953	UGAUGGAGGGCUGGGUGGCug	GCCACCCAGCCCUCCAUCA	-3.4	-2.1	0.2	0	4	4	-47.3	38.9	8.1	5	Ib	68.4	56.6	60.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1328	NM_053656	Huesken	P2rx2	22953	UGGAGGGCUGGGUGGCUGAuc	UCAGCCACCCAGCCCUCCA	-2.4	-2.1	0	-1	-1	4	-48.3	38.3	-3.8	5	II	68.4	45.2	51.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1329	NM_053656	Huesken	P2rx2	22953	GGUGGCUGAUCCUGGGAGUag	ACUCCCAGGAUCAGCCACC	-2.2	-3.3	-0.9	0	-1	3	-45.2	38.2	-8.7	1	III	63.2	27.5	44.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1330	NM_053656	Huesken	P2rx2	22953	GCUGAUCCUGGGAGUAGUGac	CACUACUCCCAGGAUCAGC	-2.1	-3.4	0.1	-1	1	3	-42	39.4	0.1	1	II	57.9	34.6	46.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1331	NM_053656	Huesken	P2rx2	22953	GAUCCUGGGAGUAGUGACUag	AGUCACUACUCCCAGGAUC	-2.1	-2.4	0.1	1	0	3	-41.1	55.1	-3.5	4	III	52.6	41.7	64.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1332	NM_053656	Huesken	P2rx2	22953	CUGGCCCAAGACAAGGGCAag	UGCCCUUGUCUUGGGCCAG	-2.1	-2.1	-6.9	-1	-2	5	-44.4	37.3	-8.4	-1	III	63.2	37.2	37	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1333	NM_053656	Huesken	P2rx2	22953	AAGACAAGGGCAAGGGUCAca	UGACCCUUGCCCUUGUCUU	-2.1	-0.9	-0.1	1	0	4	-41.1	44.9	-8.7	6	II	52.6	48.6	53.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1334	NM_053656	Huesken	P2rx2	22953	UCACAGGCCAUCUACUUGAgg	UCAAGUAGAUGGCCUGUGA	-2.4	-2.4	0.7	2	0	4	-39.6	58.5	-3.4	8	II	47.4	60.3	67.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1335	NM_053656	Huesken	P2rx2	22953	UUGUCGAACUUCUUAUGGCug	GCCAUAAGAAGUUCGACAA	-3.4	-0.9	1.6	1	5	3	-35.1	81	7.4	6	Ib	42.1	73.9	88.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1336	NM_053656	Huesken	P2rx2	22953	UAUGGCUGUAGAGCUUGUUuu	AACAAGCUCUACAGCCAUA	-0.9	-1.3	-1	1	2	3	-37.1	69.6	-4	6	II	42.1	61.8	76.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1337	NM_053656	Huesken	P2rx2	22953	CCAGUCACACAGGAAGGAGcc	CUCCUUCCUGUGUGACUGG	-2.1	-3.3	1	-1	0	2	-41.4	39.3	0.3	2	II	57.9	38.8	47.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1338	NM_053656	Huesken	P2rx2	22953	GGAGCCCACCCCGAUGGAGgu	CUCCAUCGGGGUGGGCUCC	-2.1	-3.3	-4.5	1	1	5	-47.8	25.2	-4.7	-1	II	73.7	27.9	38.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1339	NM_053656	Huesken	P2rx2	22953	GAGGUCAGAGCAGUGGCCAga	UGGCCACUGCUCUGACCUC	-2.1	-2.4	-0.5	0	0	4	-45.1	37.2	-3.8	2	III	63.2	38	57	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1340	NM_053656	Huesken	P2rx2	22953	GCAGUGGCCAGAUUGAUGAug	UCAUCAAUCUGGCCACUGC	-2.4	-3.4	1.4	1	-1	4	-40.6	53.6	-1.4	5	III	52.6	39.7	43.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1341	NM_053656	Huesken	P2rx2	22953	GGGAAUGAGACUGAAUUUCcc	GAAAUUCAGUCUCAUUCCC	-2.4	-3.3	1.8	1	-1	3	-35	59.9	10.8	5	II	42.1	47.2	39.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1342	NM_053656	Huesken	P2rx2	22953	GAAUGAGACUGAAUUUCCCug	GGGAAAUUCAGUCUCAUUC	-3.3	-2.4	-2.1	1	4	3	-35	64	7.1	3	II	42.1	54.3	72.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1343	NM_053656	Huesken	P2rx2	22953	GAGACUGAAUUUCCCUGCCug	GGCAGGGAAAUUCAGUCUC	-3.3	-2.4	-0.6	0	2	3	-39.4	63.9	12.8	3	II	52.6	52.5	51.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1344	NM_053656	Huesken	P2rx2	22953	CCAUGCACGAUAACAUCGAuu	UCGAUGUUAUCGUGCAUGG	-2.4	-3.3	-1.7	0	-2	2	-37	46.3	-8.3	3	III	47.4	40.8	57.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1345	NM_053656	Huesken	P2rx2	22953	AUUCGAAUCCCAUAGGCUUug	AAGCCUAUGGGAUUCGAAU	-0.9	-1.1	-0.8	3	2	3	-36.5	61.9	-6.3	5	II	42.1	55.8	76	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1346	NM_053656	Huesken	P2rx2	22953	GAUGAGAGUUCGAGUGGUGgu	CACCACUCGAACUCUCAUC	-2.1	-2.4	0.7	-1	2	2	-38.9	40.9	-2.2	4	II	52.6	41.4	64.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1347	NM_053656	Huesken	P2rx2	22953	GAGUUCGAGUGGUGGUAGUgc	ACUACCACCACUCGAACUC	-2.2	-2.4		0	-1	2	-39.9	53.8	-4	4	III	52.6	32.9	43.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1348	NM_053656	Huesken	P2rx2	22953	GUAGUGCCGUUUAUCUUGUaa	ACAAGAUAAACGGCACUAC	-2.2	-2.2	2.8	1	1	4	-35.1	63.2	-1.2	4	III	42.1	46.1	50.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1349	NM_053656	Huesken	P2rx2	22953	UAUCUUGUAAUACUUGGCAaa	UGCCAAGUAUUACAAGAUA	-2.1	-1.3	1.7	1	3	3	-32.8	86.5	1	8	II	31.6	69.4	76.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1350	NM_053656	Huesken	P2rx2	22953	AUCUUGUAAUACUUGGCAAac	UUGCCAAGUAUUACAAGAU	-0.9	-1.1	1.8	2	1	3	-32.4	84.6	-1.5	6	II	31.6	50.6	56.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1351	NM_053656	Huesken	P2rx2	22953	GGCAAACCUGAAGUUGUAGcc	CUACAACUUCAGGUUUGCC	-2.1	-3.3	-0.1	0	0	3	-36.5	53.5	7.3	3	II	47.4	33.1	57	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1352	NM_053656	Huesken	P2rx2	22953	AGCCUGAGGAGGCAGGGUCau	GACCCUGCCUCCUCAGGCU	-2.4	-2.1	-4.5	3	2	3	-47.4	40.5	9.8	3	II	68.4	50.5	61.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1353	NM_053656	Huesken	P2rx2	22953	UUGCACUCUGAUUCAGACAag	UGUCUGAAUCAGAGUGCAA	-2.1	-0.9	-1.4	2	0	2	-37.1	76.6	4.3	5	II	42.1	64.5	80.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1354	NM_053656	Huesken	P2rx2	22953	ACAAGUCCAGGUCACAGUUcc	AACUGUGACCUGGACUUGU	-0.9	-2.2	1.5	1	1	2	-38.9	58.1	1.5	4	II	47.4	48.2	38.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1355	NM_053656	Huesken	P2rx2	22953	GUCCAGGUCACAGUUCCAGuu	CUGGAACUGUGACCUGGAC	-2.1	-2.2	-0.1	2	2	2	-41.5	45.4	2.7	3	II	57.9	43.9	70.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1356	NM_053656	Huesken	P2rx2	22953	GUUCCAGUUGAUGAUGACUcc	AGUCAUCAUCAACUGGAAC	-2.1	-2.2	0.9	1	0	2	-36.1	63.3	1.4	3	III	42.1	44.9	33.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1357	NM_053656	Huesken	P2rx2	22953	AUGAUGACUCCAAUGACACcg	GUGUCAUUGGAGUCAUCAU	-2.2	-1.1	0.4	1	4	2	-36.3	69.5	4.8	6	Ia	42.1	67.3	59	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1358	NM_053656	Huesken	P2rx2	22953	GUUCUGUGAAGUUCUCUCCug	GGAGAGAACUUCACAGAAC	-3.3	-2.2	-2.1	2	2	2	-37.2	79.8	10.4	5	II	47.4	60.2	61.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1359	NM_053656	Huesken	P2rx2	22953	GUUCUCUCCUGCCUUCUCAac	UGAGAAGGCAGGAGAGAAC	-2.1	-2.2	1.7	3	0	3	-40.7	68.7	-1.4	5	II	52.6	43.2	32.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1360	NM_053656	Huesken	P2rx2	22953	UCCUGCCUUCUCAACAAUGaa	CAUUGUUGAGAAGGCAGGA	-2.1	-2.4	-1.5	0	1	3	-37.9	61.6	-4.9	3	II	47.4	60.1	61.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1361	NM_053656	Huesken	P2rx2	22953	CAACAAUGAAACCCAGCCUga	AGGCUGGGUUUCAUUGUUG	-2.1	-2.1	1.1	1	0	3	-37.4	56.4	-0.9	5	II	47.4	50.6	64.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1362	NM_053656	Huesken	P2rx2	22953	AUGGGUCAGAGUCCUGAUCaa	GAUCAGGACUCUGACCCAU	-2.4	-1.1	-2.5	2	2	3	-41.1	62.4	12.8	3	II	52.6	58.9	81.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1363	NM_053656	Huesken	P2rx2	22953	AUGUGCAAUGCUUGAGGUAgu	UACCUCAAGCAUUGCACAU	-1.3	-1.1	0.4	0	0	2	-36.9	59.9	-3.1	5	II	42.1	46	72.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1364	NM_053656	Huesken	P2rx2	22953	UCACUCUUCUGGCUUGCAAug	UUGCAAGCCAGAAGAGUGA	-0.9	-2.4	1.7	3	1	3	-39	69.8	1	6	II	47.4	49	72.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1365	NM_053656	Huesken	P2rx2	22953	GAACUUGAACUUGGGGUAGug	CUACCCCAAGUUCAAGUUC	-2.1	-2.4	2.2	0	1	4	-36.7	53.7	3	5	II	47.4	42.7	62.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1366	NM_053656	Huesken	P2rx2	22953	UGGAUGCUGUUCUUGAUGAgg	UCAUCAAGAACAGCAUCCA	-2.4	-2.1	1.4	1	0	2	-37.2	73	-3.8	6	II	42.1	54.6	53.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1367	NM_053656	Huesken	P2rx2	22953	GAUGAGGAUGGUGAAAUUUgg	AAAUUUCACCAUCCUCAUC	-0.9	-2.4	4.3	0	-1	2	-33.7	57.3	1.8	5	III	36.8	43.4	23.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1368	NM_053656	Huesken	P2rx2	22953	CCCAGAAAAUGGUUGUCAGaa	CUGACAACCAUUUUCUGGG	-2.1	-3.3	-1.9	-1	0	3	-36.4	58.2	0.3	1	II	47.4	36.8	40.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1369	NM_053656	Huesken	P2rx2	22953	GAAAAUGGUUGUCAGAAGUuc	ACUUCUGACAACCAUUUUC	-2.2	-2.4	1.8	0	0	2	-33.2	68	4.2	5	II	36.8	47.6	33.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1370	NM_053656	Huesken	P2rx2	22953	GUUCCAUCCUCCACCGGGCac	GCCCGGUGGAGGAUGGAAC	-3.4	-2.2	-1.5	3	3	6	-45.5	45.8	0.1	1	II	68.4	47.4	48.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1371	NM_053656	Huesken	P2rx2	22953	GACACCUCGCAGGUCUUGGag	CCAAGACCUGCGAGGUGUC	-3.3	-2.4	-2.2	2	2	3	-43	59.2	4.6	2	II	63.2	45.9	68.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1372	NM_053656	Huesken	P2rx2	22953	AUAGGGUACACAGUGCCCUgu	AGGGCACUGUGUACCCUAU	-2.1	-1.1	-3.6	3	4	4	-41.7	52.9	-4	5	II	52.6	64.9	55	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1373	NM_053656	Huesken	P2rx2	22953	AGUGCCCUGUGCGAAUCCCau	GGGAUUCGCACAGGGCACU	-3.3	-2.1	-1.2	2	4	4	-44.1	47	5.1	1	II	63.2	50.1	66	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1374	NM_053656	Huesken	P2rx2	22953	UGCGAAUCCCAUUGCCUUGca	CAAGGCAAUGGGAUUCGCA	-2.1	-2.1	-0.1	1	1	3	-39.3	65.1	8	6	Ib	52.6	58.8	64.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1375	NM_053656	Huesken	P2rx2	22953	CAUUGCCUUGCAUGUCCAGcu	CUGGACAUGCAAGGCAAUG	-2.1	-2.1	-0.4	-2	1	3	-38.8	55	-4.4	3	II	52.6	40.3	55.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1376	NM_053656	Huesken	P2rx2	22953	CCUUGCAUGUCCAGCUGUCcg	GACAGCUGGACAUGCAAGG	-2.4	-3.3	-1.4	-1	0	2	-41.4	45.4	-1.9	1	II	57.9	39.6	35.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1377	NM_053656	Huesken	P2rx2	22953	UCUGAAUGGCAGGUAGAGCug	GCUCUACCUGCCAUUCAGA	-3.4	-2.4	-0.5	2	4	3	-40.8	56.3	14.7	6	Ia	52.6	61.1	79.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1378	NM_053656	Huesken	P2rx2	22953	GAAUGGCAGGUAGAGCUGUga	ACAGCUCUACCUGCCAUUC	-2.2	-2.4	-0.5	1	0	3	-40.6	38.2	-4	4	III	52.6	40	42.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1379	NM_053656	Huesken	P2rx2	22953	UGGCAGGUAGAGCUGUGAAcc	UUCACAGCUCUACCUGCCA	-0.9	-2.1	-0.5	1	0	3	-41.6	53.3	6	5	II	52.6	43.4	51.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1380	NM_053656	Huesken	P2rx2	22953	AUGUUCCCAAGGUCUGGGAag	UCCCAGACCUUGGGAACAU	-2.4	-1.1	-5.8	1	1	3	-41.5	66.3	-3.8	5	II	52.6	47	50.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1381	NM_053656	Huesken	P2rx2	22953	AAGGGGUAACCUCGAUCCUgg	AGGAUCGAGGUUACCCCUU	-2.1	-0.9	-1.9	1	2	4	-41	49	-0.9	3	II	52.6	50.9	76.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1382	NM_053656	Huesken	P2rx2	22953	UCGAUCCUGGUGAUGAUGCug	GCAUCAUCACCAGGAUCGA	-3.4	-2.4	0.8	1	3	2	-40.4	67.4	7.1	6	II	52.6	66.1	78.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1383	NM_053656	Huesken	P2rx2	22953	GAUGAUGCUGACUACACUGcc	CAGUGUAGUCAGCAUCAUC	-2.1	-2.4	-0.8	0	2	2	-37.5	62.3	2.7	4	II	47.4	55.5	78.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1384	NM_053656	Huesken	P2rx2	22953	GGGGCUUUACGUAUUCCUCca	GAGGAAUACGUAAAGCCCC	-2.4	-3.3	0.9	1	1	5	-38.8	66	10.2	1	II	52.6	46.4	34.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1385	NM_053656	Huesken	P2rx2	22953	GUCCCACACUUUGUCUUCCga	GGAAGACAAAGUGUGGGAC	-3.3	-2.2	1	2	2	3	-39.1	59.9	12.8	2	II	52.6	46.2	69.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1386	NM_053656	Huesken	P2rx2	22953	UCCCACACUUUGUCUUCCGac	CGGAAGACAAAGUGUGGGA	-2.4	-2.4	1	1	3	3	-39.3	63	2.7	5	II	52.6	62.8	64.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1387	NM_053656	Huesken	P2rx2	22953	UGGUGAUCCCCUUGACUUUgg	AAAGUCAAGGGGAUCACCA	-0.9	-2.1	-2.2	1	-1	4	-39.1	65.1	-3.3	5	II	47.4	50.8	66.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1388	NM_053656	Huesken	P2rx2	22953	CUUGACUUUGGUGAUGAUGga	CAUCAUCACCAAAGUCAAG	-2.1	-2.1	2.5	0	0	2	-34.5	69.7	3.7	4	II	42.1	54.6	71.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1389	NM_053656	Huesken	P2rx2	22953	UGAUGAUGGAGCUCUCCGGuc	CCGGAGAGCUCCAUCAUCA	-3.3	-2.1	-3.3	2	4	4	-42.5	62.7	-2.4	5	Ia	57.9	65.6	54.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1390	NM_053656	Huesken	P2rx2	22953	CCGGUCUCGCUGUCCUGGUag	ACCAGGACAGCGAGACCGG	-2.2	-3.3	0.1	1	-2	4	-45.7	40.2	-6	1	III	68.4	36.1	49.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1391	NM_053656	Huesken	P2rx2	22953	GUCUCGCUGUCCUGGUAGCuu	GCUACCAGGACAGCGAGAC	-3.4	-2.2	-1.9	0	1	3	-43.5	49.9	2.7	0	II	63.2	37.1	60.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1392	NM_053656	Huesken	P2rx2	22953	CUGGUAGCUUUUCUGCACGau	CGUGCAGAAAAGCUACCAG	-2.4	-2.1	-0.7	-1	1	2	-38	59.3	5.8	3	II	52.6	54.8	80.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1393	NM_053656	Huesken	P2rx2	22953	UGGUAGCUUUUCUGCACGAug	UCGUGCAGAAAAGCUACCA	-2.4	-2.1	-0.7	-1	0	2	-38.3	56.1	-3.4	5	II	47.4	48.7	74.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1394	NM_053656	Huesken	P2rx2	22953	UUCUGCACGAUGAAGACGUac	ACGUCUUCAUCGUGCAGAA	-2.2	-0.9	-0.4	1	2	2	-37.8	64.8	-6.6	7	II	47.4	68	85.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1395	NM_053656	Huesken	P2rx2	22953	GAAGUAAAGCAGGAUGAGAag	UCUCAUCCUGCUUUACUUC	-2.4	-2.4	3.3	1	-1	2	-36.2	48.8	-1.5	4	II	42.1	38.9	61.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1396	NM_053656	Huesken	P2rx2	22953	CAGGCGCCGAUUCCGCACCac	GGUGCGGAAUCGGCGCCUG	-3.3	-2.1	-7.5	0	0	7	-45.9	41.4	10.8	0	II	73.7	50.6	69.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1397	NM_003348	Huesken	UBE2N	7334	UAGCCUAGUCCAUGCUCUAgc	UAGAGCAUGGACUAGGCUA	-1.3	-1.3	-1.1	0	0	3	-40.1	59.2	-6.1	4	II	47.4	54.3	73.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1398	NM_003348	Huesken	UBE2N	7334	UCCAUGCUCUAGCUGUUUCua	GAAACAGCUAGAGCAUGGA	-2.4	-2.4	-1.4	1	2	2	-38.4	77.8	14.4	5	II	47.4	59.8	100.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1399	NM_003348	Huesken	UBE2N	7334	AUGCUCUAGCUGUUUCUAUgg	AUAGAAACAGCUAGAGCAU	-1.1	-1.1	-1.4	4	1	2	-35.1	78	-1.6	5	II	36.8	54.8	91.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1400	NM_003348	Huesken	UBE2N	7334	UGCUCUAGCUGUUUCUAUGgc	CAUAGAAACAGCUAGAGCA	-2.1	-2.1	-1.4	1	2	2	-36.1	90	5.4	4	Ib	42.1	63.6	95.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1401	NM_003348	Huesken	UBE2N	7334	CUAGCUGUUUCUAUGGCUUgg	AAGCCAUAGAAACAGCUAG	-0.9	-2.1	-0.5	-1	0	3	-35.8	67.6	-5.6	3	III	42.1	49.3	90.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1402	NM_003348	Huesken	UBE2N	7334	GCUGUUUCUAUGGCUUGGGcu	CCCAAGCCAUAGAAACAGC	-3.3	-3.4	1.9	1	2	3	-39	58.8	0	4	II	52.6	48.5	60.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1403	NM_003348	Huesken	UBE2N	7334	UGUUUCUAUGGCUUGGGCUuc	AGCCCAAGCCAUAGAAACA	-2.1	-2.1	0.2	1	3	4	-39	82.5	-4	7	II	47.4	63.3	91.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1404	NM_003348	Huesken	UBE2N	7334	GUUUCUAUGGCUUGGGCUUcg	AAGCCCAAGCCAUAGAAAC	-0.9	-2.2	0.2	0	0	4	-37.8	68.4	-3.3	4	II	47.4	42	61.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1405	NM_003348	Huesken	UBE2N	7334	UUUCUAUGGCUUGGGCUUCgu	GAAGCCCAAGCCAUAGAAA	-2.4	-0.9	0.2	2	3	4	-38	66.6	10.4	8	Ia	47.4	66.9	90.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1406	NM_003348	Huesken	UBE2N	7334	UUCUAUGGCUUGGGCUUCGuu	CGAAGCCCAAGCCAUAGAA	-2.4	-0.9	0.2	0	3	4	-39.5	61.8	2.7	8	Ia	52.6	66.8	84.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1407	NM_003348	Huesken	UBE2N	7334	UCUAUGGCUUGGGCUUCGUug	ACGAAGCCCAAGCCAUAGA	-2.2	-2.4	0.2	0	2	4	-40.8	58.8	-4	7	II	52.6	53.5	77.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1408	NM_003348	Huesken	UBE2N	7334	UAUGGCUUGGGCUUCGUUGgu	CAACGAAGCCCAAGCCAUA	-2.1	-1.3	0.2	2	4	4	-39.3	77.3	0	7	Ib	52.6	71.1	94	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1409	NM_003348	Huesken	UBE2N	7334	UUGGGCUUCGUUGGUCUUCca	GAAGACCAACGAAGCCCAA	-2.4	-0.9		1	2	4	-39.3	64.1	10.4	6	II	52.6	68.5	85.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1410	NM_003348	Huesken	UBE2N	7334	CUUCGUUGGUCUUCCACUGcu	CAGUGGAAGACCAACGAAG	-2.1	-2.1	-1.4	2	1	2	-38.1	63.6	-1.3	2	II	52.6	53.6	78	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1411	NM_003348	Huesken	UBE2N	7334	GGUCUUCCACUGCUCCGCUac	AGCGGAGCAGUGGAAGACC	-2.1	-3.3	0.2	1	1	4	-44.2	46.7	0.8	1	III	63.2	35.3	43.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1412	NM_003348	Huesken	UBE2N	7334	GUCUUCCACUGCUCCGCUAca	UAGCGGAGCAGUGGAAGAC	-1.3	-2.2	0.2	1	-2	4	-42.2	54.5	-6.1	2	III	57.9	30.4	40.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1413	NM_003348	Huesken	UBE2N	7334	CUGCUCCGCUACAUCAUUUgc	AAAUGAUGUAGCGGAGCAG	-0.9	-2.1	-0.3	1	-2	4	-37.4	61.8	-6.3	3	III	47.4	43.4	43.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1414	NM_003348	Huesken	UBE2N	7334	UGCUCCGCUACAUCAUUUGcu	CAAAUGAUGUAGCGGAGCA	-2.1	-2.1	0.4	0	1	4	-37.4	75.3	-2.4	7	II	47.4	61.3	64.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1415	NM_003348	Huesken	UBE2N	7334	GCUCCGCUACAUCAUUUGCua	GCAAAUGAUGUAGCGGAGC	-3.4	-3.4	-0.1	1	1	4	-38.7	61.4	15.1	2	II	52.6	39	49.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1416	NM_003348	Huesken	UBE2N	7334	CGCUACAUCAUUUGCUAAUgg	AUUAGCAAAUGAUGUAGCG	-1.1	-2.4	-0.3	-1	-4	3	-32.9	72.4	-0.3	4	III	36.8	37.7	34.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1417	NM_003348	Huesken	UBE2N	7334	GCUACAUCAUUUGCUAAUGga	CAUUAGCAAAUGAUGUAGC	-2.1	-3.4	-1.6	1	0	2	-32.6	70.6	3	2	II	36.8	43.7	30.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1418	NM_003348	Huesken	UBE2N	7334	CUACAUCAUUUGCUAAUGGau	CCAUUAGCAAAUGAUGUAG	-3.3	-2.1	-1.9	0	2	2	-32.5	67.9	5.4	3	II	36.8	55.3	77.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1419	NM_003348	Huesken	UBE2N	7334	CAUCAUUUGCUAAUGGAUCau	GAUCCAUUAGCAAAUGAUG	-2.4	-2.1	-1.3	2	1	2	-32.8	73.4	7.8	3	II	36.8	59.9	61.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1420	NM_003348	Huesken	UBE2N	7334	AUCAUUUGCUAAUGGAUCAuc	UGAUCCAUUAGCAAAUGAU	-2.1	-1.1	-0.4	2	0	2	-32.8	84.2	-1.5	7	II	31.6	59	70.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1421	NM_003348	Huesken	UBE2N	7334	UGCUAAUGGAUCAUCUGGAuu	UCCAGAUGAUCCAUUAGCA	-2.4	-2.1	0.7	2	1	2	-37.7	67	-3.7	7	II	42.1	56.3	85.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1422	NM_003348	Huesken	UBE2N	7334	UGGAUCAUCUGGAUUGGGAgc	UCCCAAUCCAGAUGAUCCA	-2.4	-2.1	1.4	1	1	3	-39.9	69.5	-4	6	II	47.4	53.9	78.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1423	NM_003348	Huesken	UBE2N	7334	GGAUCAUCUGGAUUGGGAGca	CUCCCAAUCCAGAUGAUCC	-2.1	-3.3	1.4	1	2	3	-39.9	61.5	2.3	3	II	52.6	39.9	58.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1424	NM_003348	Huesken	UBE2N	7334	GAUAGCAGAACUGUGCGGAuc	UCCGCACAGUUCUGCUAUC	-2.4	-2.4	-1.8	0	1	4	-40	55.3	-0.8	4	II	52.6	39.6	63.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1425	NM_003348	Huesken	UBE2N	7334	AUAGCAGAACUGUGCGGAUcu	AUCCGCACAGUUCUGCUAU	-1.1	-1.1	-1.8	0	2	4	-38.7	48.9	-4	4	II	47.4	45.3	54.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1426	NM_003348	Huesken	UBE2N	7334	CAGAACUGUGCGGAUCUGCag	GCAGAUCCGCACAGUUCUG	-3.4	-2.1	-1.4	0	0	4	-40.8	51.3	5.8	3	II	57.9	53.4	86.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1427	NM_003348	Huesken	UBE2N	7334	AUCUGCAGUGCUGGGGACCac	GGUCCCCAGCACUGCAGAU	-3.3	-1.1	0.4	0	3	4	-45	48.9	8.1	4	Ib	63.2	57.1	62.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1428	NM_003348	Huesken	UBE2N	7334	CUGCAGUGCUGGGGACCACuu	GUGGUCCCCAGCACUGCAG	-2.2	-2.1	-2.4	1	0	4	-45.8	28.5	8.1	1	II	68.4	38.6	59.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1429	NM_003348	Huesken	UBE2N	7334	GCAGUGCUGGGGACCACUUau	AAGUGGUCCCCAGCACUGC	-0.9	-3.4	-2.7	1	-1	4	-44.6	33.2	-6.6	0	III	63.2	33.1	48.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1430	NM_003348	Huesken	UBE2N	7334	GGGGACCACUUAUCUUUCAaa	UGAAAGAUAAGUGGUCCCC	-2.1	-3.3	0.2	1	-3	4	-38.3	50.6	-1.1	2	III	47.4	30.6	34.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1431	NM_003348	Huesken	UBE2N	7334	GACCACUUAUCUUUCAAAAua	UUUUGAAAGAUAAGUGGUC	-0.9	-2.4	1	2	-2	2	-31.1	79	2	3	III	31.6	36.1	35.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1432	NM_003348	Huesken	UBE2N	7334	CUUAUCUUUCAAAAUAUCUag	AGAUAUUUUGAAAGAUAAG	-2.1	-2.1	0.4	0	-1	1	-26.9	93.9	-6.3	6	II	21.1	62.7	57.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1433	NM_003348	Huesken	UBE2N	7334	UUAUCUUUCAAAAUAUCUAga	UAGAUAUUUUGAAAGAUAA	-1.3	-0.9	0.4	2	2	1	-26.1	105.8	-1.7	8	II	15.8	75.2	80.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1434	NM_003348	Huesken	UBE2N	7334	UAUCUUUCAAAAUAUCUAGac	CUAGAUAUUUUGAAAGAUA	-2.1	-1.3	1.2	2	3	1	-27.3	105.7	2.3	9	Ia	21.1	78.9	85.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1435	NM_003348	Huesken	UBE2N	7334	UUUCAAAAUAUCUAGACAUau	AUGUCUAGAUAUUUUGAAA	-1.1	-0.9	1.3	1	2	1	-28.2	85.2	-3.6	8	II	21.1	70.6	75.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1436	NM_003348	Huesken	UBE2N	7334	CUAGACAUAUUCUUCCCAAcu	UUGGGAAGAAUAUGUCUAG	-0.9	-2.1	-1.2	-1	-1	3	-34	66.1	-3.1	3	II	36.8	45.1	52.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1437	NM_003348	Huesken	UBE2N	7334	UUCUUCCCAACUUGUCUACau	GUAGACAAGUUGGGAAGAA	-2.2	-0.9	2	-1	2	3	-35.8	86.6	8.1	8	Ib	42.1	62.4	94	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1438	NM_003348	Huesken	UBE2N	7334	UUCCCAACUUGUCUACAUUag	AAUGUAGACAAGUUGGGAA	-0.9	-0.9	1	1	1	3	-34.5	78.5	3.8	5	II	36.8	58.3	93.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1439	NM_003348	Huesken	UBE2N	7334	CCAACUUGUCUACAUUAGGau	CCUAAUGUAGACAAGUUGG	-3.3	-3.3	-0.9	0	0	2	-34.6	65	0.4	3	II	42.1	53.2	69.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1440	NM_003348	Huesken	UBE2N	7334	UUGUCUACAUUAGGAUGAUaa	AUCAUCCUAAUGUAGACAA	-1.1	-0.9	1	-1	1	2	-33.1	71.2	-4	6	II	31.6	55.2	46.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1441	NM_003348	Huesken	UBE2N	7334	CUACAUUAGGAUGAUAAAUuu	AUUUAUCAUCCUAAUGUAG	-1.1	-2.1	1.5	2	-2	2	-29.7	73.6	2.1	4	II	26.3	50.3	49.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1442	NM_003348	Huesken	UBE2N	7334	UAGGAUGAUAAAUUUUGGUca	ACCAAAAUUUAUCAUCCUA	-2.2	-1.3	4.1	0	3	2	-30.3	88.2	3.4	7	II	26.3	70.5	81.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1443	NM_003348	Huesken	UBE2N	7334	AAAUUUUGGUCAUGAAACGua	CGUUUCAUGACCAAAAUUU	-2.4	-0.9	0.2	2	4	2	-29.7	80.2	-2.4	7	Ia	31.6	68.8	100	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1444	NM_003348	Huesken	UBE2N	7334	AAUUUUGGUCAUGAAACGUac	ACGUUUCAUGACCAAAAUU	-2.2	-0.9	0.2	1	3	2	-31	79.4	1.4	7	II	31.6	61	92.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1445	NM_003348	Huesken	UBE2N	7334	GUCAUGAAACGUACUUUAGgg	CUAAAGUACGUUUCAUGAC	-2.1	-2.2	1	0	0	2	-31.7	76	0.4	5	II	36.8	48	79.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1446	NM_003348	Huesken	UBE2N	7334	UUAGGGGCUGCCAUUGGGUau	ACCCAAUGGCAGCCCCUAA	-2.2	-0.9	0.4	0	4	5	-43.5	48.9	-8.9	5	II	57.9	63.4	64	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1447	NM_003348	Huesken	UBE2N	7334	GGCUGCCAUUGGGUAUUCUuc	AGAAUACCCAAUGGCAGCC	-2.1	-3.3	0.6	0	-1	3	-40.4	50.9	-1.6	2	III	52.6	27.3	44.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1448	NM_003348	Huesken	UBE2N	7334	GGGUAUUCUUCUGGAAGGAau	UCCUUCCAGAAGAAUACCC	-2.4	-3.3	0	0	-2	3	-38.5	49.2	-0.4	3	II	47.4	28.6	36.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1449	NM_003348	Huesken	UBE2N	7334	UAUUCUUCUGGAAGGAAUAgu	UAUUCCUUCCAGAAGAAUA	-1.3	-1.3	0	0	0	2	-33	82.6	-6.3	7	II	31.6	62.2	83	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1450	NM_003348	Huesken	UBE2N	7334	CUUCUGGAAGGAAUAGUUCaa	GAACUAUUCCUUCCAGAAG	-2.4	-2.1	0.5	0	1	2	-34.9	63.4	5.1	5	II	42.1	55.3	87.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1451	NM_003348	Huesken	UBE2N	7334	CUGGAAGGAAUAGUUCAAGuu	CUUGAACUAUUCCUUCCAG	-2.1	-2.1	1.6	-1	0	2	-34.6	52.3	3	4	II	42.1	48.8	84.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1452	NM_003348	Huesken	UBE2N	7334	UGGAAGGAAUAGUUCAAGUuu	ACUUGAACUAUUCCUUCCA	-2.2	-2.1	1.6	0	1	2	-34.7	73.4	3.4	6	II	36.8	59.8	96.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1453	NM_003348	Huesken	UBE2N	7334	GAAGGAAUAGUUCAAGUUUaa	AAACUUGAACUAUUCCUUC	-0.9	-2.4	2.1	0	-1	2	-31.1	63.6	1.5	4	II	31.6	46.1	43.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1454	NM_003348	Huesken	UBE2N	7334	GGAAUAGUUCAAGUUUAAAag	UUUAAACUUGAACUAUUCC	-0.9	-3.3	2.1	0	-2	2	-28.8	69.2	0.9	4	II	26.3	32.8	29.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1455	NM_003348	Huesken	UBE2N	7334	AGUUCAAGUUUAAAAGUCCcu	GGACUUUUAAACUUGAACU	-3.3	-2.1	1.9	1	3	2	-30.7	72.7	4.8	6	Ia	31.6	64.2	87.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1456	NM_003348	Huesken	UBE2N	7334	AAAAGUCCCUCCCUCAAAGgg	CUUUGAGGGAGGGACUUUU	-2.1	-0.9	-0.8	2	4	3	-37.6	67.9	0	5	Ia	47.4	59.2	70.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1457	NM_003348	Huesken	UBE2N	7334	GAAUCCUGAGGGCCAGCAAug	UUGCUGGCCCUCAGGAUUC	-0.9	-2.4	-1.3	1	-1	5	-42.7	50.3	-1.4	2	II	57.9	30.5	50.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1458	NM_003348	Huesken	UBE2N	7334	GAGGGCCAGCAAUGACCACau	GUGGUCAUUGCUGGCCCUC	-2.2	-2.4	-4.9	0	1	5	-43.8	32.3	9.7	-1	II	63.2	33.1	54.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1459	NM_003348	Huesken	UBE2N	7334	GGGCCAGCAAUGACCACAUga	AUGUGGUCAUUGCUGGCCC	-1.1	-3.3	-4.9	1	-2	5	-42.5	37.7	-6.6	0	III	57.9	31.3	47.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1460	NM_003348	Huesken	UBE2N	7334	GCCAGCAAUGACCACAUGAaa	UCAUGUGGUCAUUGCUGGC	-2.4	-3.4	-0.8	0	-3	3	-40.4	47.3	-1.4	3	III	52.6	34.3	50.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1461	NM_003348	Huesken	UBE2N	7334	CCAGCAAUGACCACAUGAAaa	UUCAUGUGGUCAUUGCUGG	-0.9	-3.3	0.6	0	-3	2	-37.9	44.5	-13	3	III	47.4	32	39.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1462	NM_003348	Huesken	UBE2N	7334	CAGCAAUGACCACAUGAAAau	UUUCAUGUGGUCAUUGCUG	-0.9	-2.1	0.6	1	-4	2	-35.5	48.5	-3	3	II	42.1	32.7	39.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1463	NM_003348	Huesken	UBE2N	7334	GCAAUGACCACAUGAAAAUaa	AUUUUCAUGUGGUCAUUGC	-1.1	-3.4	1	0	-2	2	-33.3	58.5	-6.3	5	II	36.8	39.2	47.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1464	NM_003348	Huesken	UBE2N	7334	ACAUGAAAAUAACGGGCGUug	ACGCCCGUUAUUUUCAUGU	-2.2	-2.2	1.8	0	2	6	-35.1	51.1	0.7	5	II	42.1	44.2	62.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1465	NM_003348	Huesken	UBE2N	7334	AUGAAAAUAACGGGCGUUGcu	CAACGCCCGUUAUUUUCAU	-2.1	-1.1	0	0	2	6	-33.8	62.3	0.4	8	Ia	42.1	61	87.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1466	NM_003348	Huesken	UBE2N	7334	GGGCGUUGCUCUCAUCUGGuu	CCAGAUGAGAGCAACGCCC	-3.3	-3.3	1.3	3	0	5	-43.1	47.7	3.1	1	II	63.2	41	65.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1467	NM_003348	Huesken	UBE2N	7334	UGCUCUCAUCUGGUUCGGCuu	GCCGAACCAGAUGAGAGCA	-3.4	-2.1	0.8	0	4	4	-42.2	61	9.7	3	II	57.9	57.6	82.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1468	NM_003348	Huesken	UBE2N	7334	CUCUCAUCUGGUUCGGCUUug	AAGCCGAACCAGAUGAGAG	-0.9	-2.1	3.3	0	-2	4	-39.7	67.7	-3	3	II	52.6	42.4	88.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1469	NM_003348	Huesken	UBE2N	7334	AUCUGGUUCGGCUUUGAUGcc	CAUCAAAGCCGAACCAGAU	-2.1	-1.1	0.9	3	3	4	-37.2	78	0	5	Ib	47.4	62.9	84.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1470	NM_003348	Huesken	UBE2N	7334	UGGUUCGGCUUUGAUGCCAgg	UGGCAUCAAAGCCGAACCA	-2.1	-2.1	-4.5	1	0	4	-40.4	57.4	-0.7	5	II	52.6	53	81.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1471	NM_003348	Huesken	UBE2N	7334	CGGCUUUGAUGCCAGGAACug	GUUCCUGGCAUCAAAGCCG	-2.2	-2.4	-4.5	1	-2	4	-40.4	52.2	5.5	0	II	57.9	40.4	78.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1472	NM_003348	Huesken	UBE2N	7334	GCUUUGAUGCCAGGAACUGgu	CAGUUCCUGGCAUCAAAGC	-2.1	-3.4	0.2	0	1	3	-38.9	42.8	-2.4	4	II	52.6	38.6	61.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1473	NM_003348	Huesken	UBE2N	7334	UUUGAUGCCAGGAACUGGUuc	ACCAGUUCCUGGCAUCAAA	-2.2	-0.9	-2.7	1	3	3	-38.9	67.3	-3.9	7	II	47.4	69.2	97.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1474	NM_003348	Huesken	UBE2N	7334	UGAUGCCAGGAACUGGUUCug	GAACCAGUUCCUGGCAUCA	-2.4	-2.1	-2.7	1	3	3	-40.4	64.4	14.4	5	II	52.6	60.4	36	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1475	NM_003348	Huesken	UBE2N	7334	GAUGCCAGGAACUGGUUCUgc	AGAACCAGUUCCUGGCAUC	-2.1	-2.4	-2.7	0	0	3	-40.4	61.1	-4	4	III	52.6	43.9	60.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1476	NM_003342	Huesken	UBE2G1	7326	UUUUCUUACACAGCGGGCAac	UGCCCGCUGUGUAAGAAAA	-2.1	-0.9	0.8	2	2	6	-38	74.7	-6.1	8	II	47.4	66.8	94.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1477	NM_003342	Huesken	UBE2G1	7326	CUUACACAGCGGGCAACUUuu	AAGUUGCCCGCUGUGUAAG	-0.9	-2.1	0.2	0	-1	6	-39	47.2	-6	3	II	52.6	36.2	79	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1478	NM_003342	Huesken	UBE2G1	7326	UUACACAGCGGGCAACUUUuc	AAAGUUGCCCGCUGUGUAA	-0.9	-0.9	0.4	2	1	6	-37.8	61	1.1	8	II	47.4	63.3	93.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1479	NM_003342	Huesken	UBE2G1	7326	GGGCAACUUUUCUUUUAAAuu	UUUAAAAGAAAAGUUGCCC	-0.9	-3.3	2.6	1	-3	4	-30.3	63	-1.1	1	III	31.6	24.5	43.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1480	NM_003342	Huesken	UBE2G1	7326	ACUUUUCUUUUAAAUUCUCca	GAGAAUUUAAAAGAAAAGU	-2.4	-2.2	2.3	1	3	1	-26.2	87.1	4.8	5	Ia	21.1	62.6	56.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1481	NM_003342	Huesken	UBE2G1	7326	UUUUCUUUUAAAUUCUCCAuu	UGGAGAAUUUAAAAGAAAA	-2.1	-0.9	2.3	1	3	2	-27.3	113.4	0.9	8	II	21.1	78.6	74.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1482	NM_003342	Huesken	UBE2G1	7326	AAAUUCUCCAUUUCUAUCUuc	AGAUAGAAAUGGAGAAUUU	-2.1	-0.9	0.9	3	2	2	-30.4	103.9	-1	9	II	26.3	73.7	102	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1483	NM_003342	Huesken	UBE2G1	7326	UUUCUAUCUUCCCUCCAUUcu	AAUGGAGGGAAGAUAGAAA	-0.9	-0.9	3.9	2	2	3	-35	81.4	-3.9	7	II	36.8	65.4	85.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1484	NM_003342	Huesken	UBE2G1	7326	UUCCCUCCAUUCUUUCGCAgc	UGCGAAAGAAUGGAGGGAA	-2.1	-0.9	2.7	1	2	3	-38.4	71.6	-3.8	4	II	47.4	56.5	72.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1485	NM_003342	Huesken	UBE2G1	7326	CCCUCCAUUCUUUCGCAGCau	GCUGCGAAAGAAUGGAGGG	-3.4	-3.3	0.8	-1	-1	3	-40.6	63.4	8.4	1	II	57.9	46.4	63.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1486	NM_003342	Huesken	UBE2G1	7326	CCUCCAUUCUUUCGCAGCAuc	UGCUGCGAAAGAAUGGAGG	-2.1	-3.3	0.8	1	-2	3	-39.4	51.1	-0.3	2	III	52.6	34.1	34.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1487	NM_003342	Huesken	UBE2G1	7326	UCUUUCGCAGCAUCAACAUua	AUGUUGAUGCUGCGAAAGA	-1.1	-2.4	0.6	0	1	3	-36.1	70.4	-6.3	7	II	42.1	53.7	104.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1488	NM_003342	Huesken	UBE2G1	7326	CUUUCGCAGCAUCAACAUUag	AAUGUUGAUGCUGCGAAAG	-0.9	-2.1	0.6	0	-2	3	-34.6	64	1.8	3	III	42.1	40.8	93.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1489	NM_003342	Huesken	UBE2G1	7326	CAGCAUCAACAUUAGCAGGug	CCUGCUAAUGUUGAUGCUG	-3.3	-2.1	-0.1	0	0	2	-36.8	62.3	6	2	II	47.4	49.7	97.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1490	NM_003342	Huesken	UBE2G1	7326	AGCAUCAACAUUAGCAGGUga	ACCUGCUAAUGUUGAUGCU	-2.2	-2.1	-0.1	1	1	2	-36.9	64.8	-3.5	6	II	42.1	54.4	93.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1491	NM_003342	Huesken	UBE2G1	7326	AACAUUAGCAGGUGAGUCUcc	AGACUCACCUGCUAAUGUU	-2.1	-0.9	1.3	1	1	2	-37	75.3	-4	8	II	42.1	60.1	81.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1492	NM_003342	Huesken	UBE2G1	7326	GCAGGUGAGUCUCCAUUAGgg	CUAAUGGAGACUCACCUGC	-2.1	-3.4	0.3	1	0	2	-39.6	46.8	3.1	3	II	52.6	35.3	71.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1493	NM_003342	Huesken	UBE2G1	7326	UGAGUCUCCAUUAGGGUCUgc	AGACCCUAAUGGAGACUCA	-2.1	-2.1	-1.1	1	1	3	-39.8	70.2	-3.5	7	II	47.4	66.6	89.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1494	NM_003342	Huesken	UBE2G1	7326	AGGGUCUGCCAGCAUAGAAau	UUCUAUGCUGGCAGACCCU	-0.9	-2.1	0.4	3	-1	3	-41.7	53.1	1	4	II	52.6	39.6	61	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1495	NM_003342	Huesken	UBE2G1	7326	CCAGCAUAGAAAUGACACUaa	AGUGUCAUUUCUAUGCUGG	-2.1	-3.3	0	0	-2	2	-35.9	53.3	-3.7	3	III	42.1	43.1	47.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1496	NM_003342	Huesken	UBE2G1	7326	AGCAUAGAAAUGACACUAAuc	UUAGUGUCAUUUCUAUGCU	-0.9	-2.1	0.4	1	-2	2	-32.7	64.2	-6.3	7	II	31.6	44.3	54	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1497	NM_003342	Huesken	UBE2G1	7326	AUAGAAAUGACACUAAUCAug	UGAUUAGUGUCAUUUCUAU	-2.1	-1.1	-1.2	2	2	1	-30.7	75.7	1.3	7	II	26.3	64.6	78.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1498	NM_003342	Huesken	UBE2G1	7326	GACACUAAUCAUGAUGGUUuc	AACCAUCAUGAUUAGUGUC	-0.9	-2.4	0.6	0	-1	2	-34	69.5	1.4	3	II	36.8	42.2	65.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1499	NM_003342	Huesken	UBE2G1	7326	ACACUAAUCAUGAUGGUUUcc	AAACCAUCAUGAUUAGUGU	-0.9	-2.2	0.6	3	1	2	-32.5	74.1	-4.2	7	II	31.6	58.5	66.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1500	NM_003342	Huesken	UBE2G1	7326	ACUAAUCAUGAUGGUUUCCac	GGAAACCAUCAUGAUUAGU	-3.3	-2.2	0.7	0	3	2	-33.9	74.1	15.4	6	Ia	36.8	58.2	54.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1501	NM_003342	Huesken	UBE2G1	7326	AUGAUGGUUUCCACAGUGUgg	ACACUGUGGAAACCAUCAU	-2.2	-1.1	-0.9	1	1	2	-36.8	64.9	-11.3	7	II	42.1	57.9	89.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1502	NM_003342	Huesken	UBE2G1	7326	GAUGGUUUCCACAGUGUGGau	CCACACUGUGGAAACCAUC	-3.3	-2.4	-3	1	2	2	-39	60.6	-2.6	3	II	52.6	50.5	86.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1503	NM_003342	Huesken	UBE2G1	7326	AUGGUUUCCACAGUGUGGAua	UCCACACUGUGGAAACCAU	-2.4	-1.1	-4.6	3	2	2	-39	65.9	-3.8	5	II	47.4	57.8	85.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1504	NM_003342	Huesken	UBE2G1	7326	GGUUUCCACAGUGUGGAUAgg	UAUCCACACUGUGGAAACC	-1.3	-3.3	-4.9	1	-2	2	-38.2	64.8	1.6	3	III	47.4	30.2	41.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1505	NM_003342	Huesken	UBE2G1	7326	UCCACAGUGUGGAUAGGGAgc	UCCCUAUCCACACUGUGGA	-2.4	-2.4	-4.1	1	1	3	-42	53	-4	4	II	52.6	44.1	62.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1506	NM_003342	Huesken	UBE2G1	7326	CCACAGUGUGGAUAGGGAGcc	CUCCCUAUCCACACUGUGG	-2.1	-3.3	-2	1	0	3	-41.7	40.3	0.7	1	II	57.9	41.5	62.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1507	NM_003342	Huesken	UBE2G1	7326	GUGUGGAUAGGGAGCCAGCgu	GCUGGCUCCCUAUCCACAC	-3.4	-2.2	-0.4	-1	1	3	-44.2	41.4	4.8	2	II	63.2	47.1	64	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1508	NM_003342	Huesken	UBE2G1	7326	GGGAGCCAGCGUUCCUCUGgc	CAGAGGAACGCUGGCUCCC	-2.1	-3.3	-1.6	1	0	3	-45.3	42.5	3	-1	II	68.4	33.2	42.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1509	NM_003342	Huesken	UBE2G1	7326	AGCCAGCGUUCCUCUGGCUuu	AGCCAGAGGAACGCUGGCU	-2.1	-2.1	-7.9	2	1	3	-45.1	39.5	-6	1	II	63.2	38.6	37.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1510	NM_003342	Huesken	UBE2G1	7326	GCCAGCGUUCCUCUGGCUUuu	AAGCCAGAGGAACGCUGGC	-0.9	-3.4	-6.5	0	-1	3	-43.9	45.4	1.4	0	III	63.2	29	35.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1511	NM_003342	Huesken	UBE2G1	7326	CAGCGUUCCUCUGGCUUUUca	AAAAGCCAGAGGAACGCUG	-0.9	-2.1	-0.5	1	-3	3	-39	52.2	-5.3	3	III	52.6	38.9	32.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1512	NM_003342	Huesken	UBE2G1	7326	GCGUUCCUCUGGCUUUUCAua	UGAAAAGCCAGAGGAACGC	-2.1	-3.4	1.1	1	-2	3	-39.3	63	1	3	III	52.6	28.5	35.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1513	NM_003342	Huesken	UBE2G1	7326	UCCUCUGGCUUUUCAUAACca	GUUAUGAAAAGCCAGAGGA	-2.2	-2.4	1	1	1	3	-35.9	79.5	10.4	5	II	42.1	58.3	49.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1514	NM_003342	Huesken	UBE2G1	7326	UCUGGCUUUUCAUAACCAUac	AUGGUUAUGAAAAGCCAGA	-1.1	-2.4	-4.2	1	1	3	-34.6	68.1	-6.3	4	II	36.8	54.6	39.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1515	NM_003342	Huesken	UBE2G1	7326	CUGGCUUUUCAUAACCAUAcu	UAUGGUUAUGAAAAGCCAG	-1.3	-2.1	-4.2	0	-4	3	-33.5	71.7	-3	1	III	36.8	49	31.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1516	NM_003342	Huesken	UBE2G1	7326	UUUUCAUAACCAUACUUAUcu	AUAAGUAUGGUUAUGAAAA	-1.1	-0.9	1.3	1	1	2	-28.1	96.3	-6.3	8	II	21.1	64.5	69.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1517	NM_003342	Huesken	UBE2G1	7326	UUUCAUAACCAUACUUAUCuu	GAUAAGUAUGGUUAUGAAA	-2.4	-0.9	1.3	2	3	2	-29.6	94.8	12.8	7	Ia	26.3	78.4	87.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1518	NM_003342	Huesken	UBE2G1	7326	CCAUACUUAUCUUCCCCAGgc	CUGGGGAAGAUAAGUAUGG	-2.1	-3.3	1	0	0	4	-37.3	68	1.4	4	II	47.4	45	55	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1519	NM_003342	Huesken	UBE2G1	7326	UACUUAUCUUCCCCAGGCUca	AGCCUGGGGAAGAUAAGUA	-2.1	-1.3	0.7	1	2	4	-39.6	71.3	-6.3	7	II	47.4	61	78.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1520	NM_003342	Huesken	UBE2G1	7326	UAUCUUCCCCAGGCUCAUGaa	CAUGAGCCUGGGGAAGAUA	-2.1	-1.3	0.7	2	3	4	-40.8	68.1	2.3	6	Ia	52.6	64.4	76	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1521	NM_003342	Huesken	UBE2G1	7326	AUCUUCCCCAGGCUCAUGAag	UCAUGAGCCUGGGGAAGAU	-2.4	-1.1	0.7	2	1	4	-41.9	72.1	1	7	II	52.6	51.2	76.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1522	NM_003342	Huesken	UBE2G1	7326	GGCUCAUGAAGAAUAGAAAug	UUUCUAUUCUUCAUGAGCC	-0.9	-3.3	0.3	1	-3	3	-33.9	63.3	-4	3	III	36.8	28.3	29.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1523	NM_003342	Huesken	UBE2G1	7326	AAUAGAAAUGCACACAUCAcc	UGAUGUGUGCAUUUCUAUU	-2.1	-0.9	-0.9	1	1	2	-32.5	69	-3.7	7	II	31.6	61.2	74.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1524	NM_003342	Huesken	UBE2G1	7326	UAGAAAUGCACACAUCACCau	GGUGAUGUGUGCAUUUCUA	-3.3	-1.3	-0.9	2	3	2	-36	74.5	10.2	8	Ia	42.1	79.3	94.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1525	NM_003342	Huesken	UBE2G1	7326	AGAAAUGCACACAUCACCAuu	UGGUGAUGUGUGCAUUUCU	-2.1	-2.1	-0.9	1	2	2	-36.8	64.9	-1.4	6	II	42.1	57.6	85	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1526	NM_003342	Huesken	UBE2G1	7326	UUUUAUCAACAUUUGGGUGcc	CACCCAAAUGUUGAUAAAA	-2.1	-0.9	1.8	0	5	3	-30.7	97.2	10.4	7	Ia	31.6	70.3	73.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1527	NM_003342	Huesken	UBE2G1	7326	UCAACAUUUGGGUGCCAGAuu	UCUGGCACCCAAAUGUUGA	-2.4	-2.4	-0.2	0	0	3	-38.9	63.3	-3.8	6	II	47.4	53.2	73.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1528	NM_003342	Huesken	UBE2G1	7326	AACAUUUGGGUGCCAGAUUuc	AAUCUGGCACCCAAAUGUU	-0.9	-0.9	-0.1	3	0	3	-36.4	68.3	-1.6	6	II	42.1	55.9	73.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1529	NM_003342	Huesken	UBE2G1	7326	GUGCCAGAUUUCUGUAAUGaa	CAUUACAGAAAUCUGGCAC	-2.1	-2.2	-2.5	0	0	3	-34.7	55.5	-2.4	1	II	42.1	43	50.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1530	NM_003342	Huesken	UBE2G1	7326	AUUUCUGUAAUGAAUUUCAuu	UGAAAUUCAUUACAGAAAU	-2.1	-1.1	0	1	1	1	-27.8	94.6	-6.3	8	II	21.1	66.1	61.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1531	NM_003342	Huesken	UBE2G1	7326	UAAUGAAUUUCAUUUUAGGag	CCUAAAAUGAAAUUCAUUA	-3.3	-1.3	-1.1	0	5	2	-26.6	96.4	0	6	Ia	21.1	75.7	69.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1532	NM_003342	Huesken	UBE2G1	7326	AUGAAUUUCAUUUUAGGAGgu	CUCCUAAAAUGAAAUUCAU	-2.1	-1.1	0.5	3	4	2	-28.9	98.1	8.1	7	Ia	26.3	71.9	87.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1533	NM_003342	Huesken	UBE2G1	7326	GAAUUUCAUUUUAGGAGGUcg	ACCUCCUAAAAUGAAAUUC	-2.2	-2.4	2.3	-1	1	2	-31.2	69.3	-3.5	4	II	31.6	43.8	39.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1534	NM_003342	Huesken	UBE2G1	7326	AUUUUAGGAGGUCGGAGGGga	CCCUCCGACCUCCUAAAAU	-3.3	-1.1		0	5	3	-39.7	62.4	5.4	7	Ia	52.6	56.3	62.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1535	NM_003342	Huesken	UBE2G1	7326	GAGGUCGGAGGGGAUAAUCuu	GAUUAUCCCCUCCGACCUC	-2.4	-2.4	1.8	0	0	4	-41.7	44.6	7.4	2	II	57.9	44.6	84.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1536	NM_003342	Huesken	UBE2G1	7326	AGGUCGGAGGGGAUAAUCUuu	AGAUUAUCCCCUCCGACCU	-2.1	-2.1	1.8	2	1	4	-41.4	54.8	-4.2	5	II	52.6	50.5	84.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1537	NM_003342	Huesken	UBE2G1	7326	CACCUUCAUAAAGUGUAUCug	GAUACACUUUAUGAAGGUG	-2.4	-2.1	-2.2	0	0	2	-32.7	73.5	10	2	II	36.8	53	66	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1538	NM_003342	Huesken	UBE2G1	7326	ACCUUCAUAAAGUGUAUCUgg	AGAUACACUUUAUGAAGGU	-2.1	-2.2	0.3	0	0	2	-32.7	86.2	-1.7	7	II	31.6	54.9	78.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1539	NM_003342	Huesken	UBE2G1	7326	UAAAGUGUAUCUGGAGGGCca	GCCCUCCAGAUACACUUUA	-3.4	-1.3	2.5	-1	5	4	-38.5	62.6	8.1	7	Ia	47.4	66.2	75.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1540	NM_003342	Huesken	UBE2G1	7326	GUGUAUCUGGAGGGCCAAUaa	AUUGGCCCUCCAGAUACAC	-1.1	-2.2	-1.5	0	-2	5	-40.7	45.7	1	2	II	52.6	30.8	36.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1541	NM_003342	Huesken	UBE2G1	7326	GGAGGGCCAAUAAUAAGGAcu	UCCUUAUUAUUGGCCCUCC	-2.4	-3.3	-0.2	0	0	5	-38.5	39.3	-4	2	III	47.4	34.9	44.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1542	NM_003342	Huesken	UBE2G1	7326	GGCCAAUAAUAAGGACUUCcc	GAAGUCCUUAUUAUUGGCC	-2.4	-3.3	-0.2	1	-1	4	-35	47.3	12.4	3	II	42.1	34.6	87.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1543	NM_003342	Huesken	UBE2G1	7326	GCCAAUAAUAAGGACUUCCca	GGAAGUCCUUAUUAUUGGC	-3.3	-3.4	-1	0	0	3	-35	66	12.4	4	II	42.1	54.6	71	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1544	NM_003342	Huesken	UBE2G1	7326	UAAGGACUUCCCAUCGGUAga	UACCGAUGGGAAGUCCUUA	-1.3	-1.3	-2	0	2	3	-39	54.8	-8.7	4	II	47.4	54.4	93.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1545	NM_003342	Huesken	UBE2G1	7326	GGACUUCCCAUCGGUAGAGau	CUCUACCGAUGGGAAGUCC	-2.1	-3.3	-0.9	1	1	3	-41.3	41.9	-2.4	2	II	57.9	36.5	68.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1546	NM_003342	Huesken	UBE2G1	7326	ACUUCCCAUCGGUAGAGAUca	AUCUCUACCGAUGGGAAGU	-1.1	-2.2	-2.7	1	1	3	-39.1	63.1	-4	3	II	47.4	38.7	67.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1547	NM_003342	Huesken	UBE2G1	7326	CCAUCGGUAGAGAUCAUUGuc	CAAUGAUCUCUACCGAUGG	-2.1	-3.3	-0.1	-1	0	3	-36.8	63.5	0	4	II	47.4	50.2	86.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1548	NM_003342	Huesken	UBE2G1	7326	AUCGGUAGAGAUCAUUGUCau	GACAAUGAUCUCUACCGAU	-2.4	-1.1	0.6	1	3	3	-36	70.8	15.1	4	Ib	42.1	62.5	103.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1549	NM_003342	Huesken	UBE2G1	7326	CGGUAGAGAUCAUUGUCAUcu	AUGACAAUGAUCUCUACCG	-1.1	-2.4	1.7	-1	-2	3	-35.7	58.5	-3.3	3	III	42.1	34.2	40.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1550	NM_003342	Huesken	UBE2G1	7326	UUGUCAUCUAUUAAACCUGca	CAGGUUUAAUAGAUGACAA	-2.1	-0.9	-0.4	1	3	2	-31.4	86	0.4	7	Ia	31.6	78.3	97.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1551	NM_003342	Huesken	UBE2G1	7326	CAUCUAUUAAACCUGCAGAaa	UCUGCAGGUUUAAUAGAUG	-2.4	-2.1	0.7	1	-1	2	-33.8	78	-1.1	5	II	36.8	48.4	60.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1552	NM_003342	Huesken	UBE2G1	7326	UCUAUUAAACCUGCAGAAAag	UUUCUGCAGGUUUAAUAGA	-0.9	-2.4	1.1	0	-1	2	-32.4	75.4	-3.1	7	II	31.6	51.9	51.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1553	NM_003342	Huesken	UBE2G1	7326	UUAAACCUGCAGAAAAGCCuu	GGCUUUUCUGCAGGUUUAA	-3.3	-0.9	-1.7	1	5	3	-35.2	71.1	9.8	6	Ia	42.1	75.9	83.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1554	NM_003342	Huesken	UBE2G1	7326	GAAAAGCCUUCCACUGGAUuu	AUCCAGUGGAAGGCUUUUC	-1.1	-2.4	-0.6	0	0	3	-37.9	48.5	-8.6	4	II	47.4	34.7	65.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1555	NM_006357	Huesken	UBE2E3	10477	CACUGUCUGGCUAUCCUGUcg	ACAGGAUAGCCAGACAGUG	-2.2	-2.1	-0.9	0	-1	3	-40.6	61.7	-5.6	3	III	52.6	43.4	55.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1556	NM_006357	Huesken	UBE2E3	10477	CUGUCUGGCUAUCCUGUCGug	CGACAGGAUAGCCAGACAG	-2.4	-2.1	-0.9	0	0	3	-41.1	68.1	5.7	3	II	57.9	52.8	78	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1557	NM_006357	Huesken	UBE2E3	10477	UGUCGUGUUCUGCUCUGUUgg	AACAGAGCAGAACACGACA	-0.9	-2.1		1	2	2	-38.3	65.3	0.8	5	II	47.4	55.9	76.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1558	NM_006357	Huesken	UBE2E3	10477	UCGUGUUCUGCUCUGUUGGuc	CCAACAGAGCAGAACACGA	-3.3	-2.4		1	3	2	-39.4	77.1	5.4	6	Ib	52.6	66.3	88.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1559	NM_006357	Huesken	UBE2E3	10477	GUUGGUCAAAUACUGAGUGgc	CACUCAGUAUUUGACCAAC	-2.1	-2.2	1	0	3	2	-34.6	64	4.7	3	II	42.1	51.8	87.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1560	NM_006357	Huesken	UBE2E3	10477	UGGUCAAAUACUGAGUGGCua	GCCACUCAGUAUUUGACCA	-3.4	-2.1	1	0	3	3	-38.2	68.4	8.1	6	Ib	47.4	65.2	90.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1561	NM_006357	Huesken	UBE2E3	10477	UCAAAUACUGAGUGGCUAUgc	AUAGCCACUCAGUAUUUGA	-1.1	-2.4	1.4	-1	0	3	-35.1	73	1	8	II	36.8	57.7	88.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1562	NM_006357	Huesken	UBE2E3	10477	GCUUCCAACCAGAGGAUCCgc	GGAUCCUCUGGUUGGAAGC	-3.3	-3.4	-2.4	0	1	2	-41.8	55.6	7.1	2	II	57.9	45.2	59.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1563	NM_006357	Huesken	UBE2E3	10477	UCCAACCAGAGGAUCCGCAgg	UGCGGAUCCUCUGGUUGGA	-2.1	-2.4	-1.6	1	1	4	-43.3	51.6	-1.3	4	II	57.9	50.8	75.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1564	NM_006357	Huesken	UBE2E3	10477	CAGAGGAUCCGCAGGGUUGca	CAACCCUGCGGAUCCUCUG	-2.1	-2.1	-0.2	-1	-1	4	-43	43.1	-6.9	3	II	63.2	46.9	56.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1565	NM_006357	Huesken	UBE2E3	10477	GGAUCCGCAGGGUUGCAGUcu	ACUGCAACCCUGCGGAUCC	-2.2	-3.3	-5.5	0	0	4	-44.1	51.8	-1.6	1	III	63.2	34	71.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1566	NM_006357	Huesken	UBE2E3	10477	GGGUUGCAGUCUGUCAAAAgg	UUUUGACAGACUGCAACCC	-0.9	-3.3	1.1	1	-3	3	-37.6	50.3	-0.8	2	III	47.4	23	50.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1567	NM_006357	Huesken	UBE2E3	10477	CUGUCAAAAGGGAACAAAUag	AUUUGUUCCCUUUUGACAG	-1.1	-2.1	-0.2	-1	-3	3	-32.8	62.8	-8.6	3	II	36.8	43.9	47.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1568	NM_006357	Huesken	UBE2E3	10477	UGUCAAAAGGGAACAAAUAga	UAUUUGUUCCCUUUUGACA	-1.3	-2.1	0.6	2	-1	3	-32	60.7	-3.6	6	II	31.6	53.5	45.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1569	NM_006357	Huesken	UBE2E3	10477	AAAAGGGAACAAAUAGACAgc	UGUCUAUUUGUUCCCUUUU	-2.1	-0.9	3.9	1	2	3	-32	64.4	-1.7	6	II	31.6	58.7	86	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1570	NM_006357	Huesken	UBE2E3	10477	AGGGAACAAAUAGACAGCAaa	UGCUGUCUAUUUGUUCCCU	-2.1	-2.1	2.1	1	0	3	-36.9	43.1	-1.1	4	II	42.1	46.8	73.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1571	NM_006357	Huesken	UBE2E3	10477	ACAAAUAGACAGCAAAACCuu	GGUUUUGCUGUCUAUUUGU	-3.3	-2.2	1.1	1	3	2	-33.1	69.6	14.8	6	Ia	36.8	67.2	94.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1572	NM_006357	Huesken	UBE2E3	10477	AUAGACAGCAAAACCUUUGaa	CAAAGGUUUUGCUGUCUAU	-2.1	-1.1	0.8	2	3	2	-33	73.2	2.4	8	Ia	36.8	71.8	92.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1573	NM_006357	Huesken	UBE2E3	10477	GCAAAACCUUUGAAAUAGUca	ACUAUUUCAAAGGUUUUGC	-2.2	-3.4	1.2	0	-1	2	-30.6	53.4	-3.9	3	II	31.6	36.7	34.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1574	NM_006357	Huesken	UBE2E3	10477	AAAUAGUCAAAGCGGGACUcc	AGUCCCGCUUUGACUAUUU	-2.1	-0.9	-1.8	1	2	5	-36	69.2	3.4	7	II	42.1	60.3	56.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1575	NM_006357	Huesken	UBE2E3	10477	AUAGUCAAAGCGGGACUCCag	GGAGUCCCGCUUUGACUAU	-3.3	-1.1	-2.2	0	4	5	-39.9	51.9	5.1	6	Ia	52.6	64.4	69.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1576	NM_006357	Huesken	UBE2E3	10477	AGUCAAAGCGGGACUCCAGuu	CUGGAGUCCCGCUUUGACU	-2.1	-2.1	-1.5	3	3	5	-41.7	45.4	0.1	3	Ia	57.9	49	72	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1577	NM_006357	Huesken	UBE2E3	10477	GGGACUCCAGUUGUCUUUAag	UAAAGACAACUGGAGUCCC	-1.3	-3.3	-1.2	0	-3	3	-38.2	59.6	1.6	1	III	47.4	30.8	43.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1578	NM_006357	Huesken	UBE2E3	10477	ACUCCAGUUGUCUUUAAGGau	CCUUAAAGACAACUGGAGU	-3.3	-2.2	0.3	2	4	2	-35.5	70.9	0	4	II	42.1	58.8	58.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1579	NM_006357	Huesken	UBE2E3	10477	GGAUGUCCAGACAGAUGACuc	GUCAUCUGUCUGGACAUCC	-2.2	-3.3	-1.3	-1	1	2	-40	46.1	4.8	1	II	52.6	33.9	43.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1580	NM_006357	Huesken	UBE2E3	10477	UGUCCAGACAGAUGACUCCcu	GGAGUCAUCUGUCUGGACA	-3.3	-2.1	-1.5	1	3	2	-41	61.8	7.4	6	II	52.6	64.1	75.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1581	NM_006357	Huesken	UBE2E3	10477	CAGACAGAUGACUCCCUGAcu	UCAGGGAGUCAUCUGUCUG	-2.4	-2.1	-1.9	-1	-3	3	-40.9	55.2	-5.8	3	III	52.6	41.8	49	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1582	NM_006357	Huesken	UBE2E3	10477	AGACAGAUGACUCCCUGACug	GUCAGGGAGUCAUCUGUCU	-2.2	-2.1	-1.9	3	3	3	-41	57.6	13.2	5	Ib	52.6	55.9	83.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1583	NM_006357	Huesken	UBE2E3	10477	AGAUGACUCCCUGACUGUUga	AACAGUCAGGGAGUCAUCU	-0.9	-2.1	0.1	2	1	3	-39.5	53.6	-3.3	4	II	47.4	42.6	58.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1584	NM_006357	Huesken	UBE2E3	10477	AUGACUCCCUGACUGUUGAug	UCAACAGUCAGGGAGUCAU	-2.4	-1.1	2.3	2	1	3	-39.5	71.1	1	5	II	47.4	49.1	74.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1585	NM_006357	Huesken	UBE2E3	10477	UGACUCCCUGACUGUUGAUgu	AUCAACAGUCAGGGAGUCA	-1.1	-2.1	2	0	1	3	-39.5	62	-4	5	II	47.4	45.7	70.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1586	NM_006357	Huesken	UBE2E3	10477	ACUCCCUGACUGUUGAUGUug	ACAUCAACAGUCAGGGAGU	-2.2	-2.2	2	3	2	3	-39.3	70.1	-1.6	5	II	47.4	50.6	69.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1587	NM_006357	Huesken	UBE2E3	10477	CCCUGACUGUUGAUGUUGCag	GCAACAUCAACAGUCAGGG	-3.4	-3.3	2	0	0	3	-39	53.4	5.1	2	II	52.6	39.2	66.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1588	NM_006357	Huesken	UBE2E3	10477	CUGUUGAUGUUGCAGUGAUag	AUCACUGCAACAUCAACAG	-1.1	-2.1	0.6	0	-2	2	-35.6	61.4	-3.6	4	II	42.1	41.8	69.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1589	NM_006357	Huesken	UBE2E3	10477	AGUGAUAGAUUCUGGUGCGga	CGCACCAGAAUCUAUCACU	-2.4	-2.1	1	0	4	3	-37.7	59.3	0.4	5	Ia	47.4	56.6	73.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1590	NM_006357	Huesken	UBE2E3	10477	GAUAGAUUCUGGUGCGGAAag	UUCCGCACCAGAAUCUAUC	-0.9	-2.4	1.5	1	-1	4	-37.9	56.5	-3.8	4	II	47.4	29.4	42.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1591	NM_006357	Huesken	UBE2E3	10477	GAUUCUGGUGCGGAAAGUAac	UACUUUCCGCACCAGAAUC	-1.3	-2.4	1.3	0	-1	4	-37.5	50.8	-6.1	4	II	47.4	35.8	49.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1592	NM_006357	Huesken	UBE2E3	10477	UCUGGUGCGGAAAGUAACCuu	GGUUACUUUCCGCACCAGA	-3.3	-2.4	-0.8	0	3	4	-39.5	53.5	7.1	5	II	52.6	66.2	91.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1593	NM_006357	Huesken	UBE2E3	10477	GGUGCGGAAAGUAACCUUUgg	AAAGGUUACUUUCCGCACC	-0.9	-3.3	-0.8	0	-2	4	-36.8	55	-3.5	4	III	47.4	39.3	50.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1594	NM_006357	Huesken	UBE2E3	10477	CGGAAAGUAACCUUUGGUGgc	CACCAAAGGUUACUUUCCG	-2.1	-2.4	-0.8	-1	0	3	-35.5	62	-1.7	4	II	47.4	45.9	58.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1595	NM_006357	Huesken	UBE2E3	10477	AACCUUUGGUGGCUUAAAUgg	AUUUAAGCCACCAAAGGUU	-1.1	-0.9	-2.3	4	1	3	-33.9	77.7	0.8	4	II	36.8	50	64.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1596	NM_006357	Huesken	UBE2E3	10477	UGGUGGCUUAAAUGGAUAAuc	UUAUCCAUUUAAGCCACCA	-0.9	-2.1	1.8	-1	-2	3	-34.7	69.3	-1.5	6	II	36.8	47.1	28.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1597	NM_006357	Huesken	UBE2E3	10477	GCUUAAAUGGAUAAUCUGAug	UCAGAUUAUCCAUUUAAGC	-2.4	-3.4	1.5	1	-2	2	-31.8	68.7	1.7	5	II	31.6	38.5	38.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1598	NM_006357	Huesken	UBE2E3	10477	CUGAUGAAAAUGUGAUAUCca	GAUAUCACAUUUUCAUCAG	-2.4	-2.1	1.1	-2	-1	1	-30.7	72.7	5.4	5	II	31.6	58.1	57.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1599	NM_006357	Huesken	UBE2E3	10477	UGAAAAUGUGAUAUCCAGAaa	UCUGGAUAUCACAUUUUCA	-2.4	-2.1	0.8	1	1	2	-32.9	77.9	4	7	II	31.6	63.3	68.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1600	NM_006357	Huesken	UBE2E3	10477	AAAACACACCACCUUCAUAua	UAUGAAGGUGGUGUGUUUU	-1.3	-0.9	2.6	2	1	2	-34.3	71.5	0.9	5	II	36.8	49.3	50.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1601	NM_006357	Huesken	UBE2E3	10477	ACACCACCUUCAUAUACAGaa	CUGUAUAUGAAGGUGGUGU	-2.1	-2.2	0.8	2	3	2	-36.1	57.3	-2.4	3	II	42.1	47.7	36.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1602	NM_006357	Huesken	UBE2E3	10477	CAGAACCCGGUGGACCAAGua	CUUGGUCCACCGGGUUCUG	-2.1	-2.1	-4.2	0	-1	5	-42.6	35.7	0.7	1	II	63.2	41.9	68.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1603	NM_006357	Huesken	UBE2E3	10477	UGGACCAAGUAUAGUUGAUcu	AUCAACUAUACUUGGUCCA	-1.1	-2.1	-0.7	0	0	2	-35.1	70.4	-1.2	5	II	36.8	47.7	59.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1604	NM_006357	Huesken	UBE2E3	10477	GACCAAGUAUAGUUGAUCUcc	AGAUCAACUAUACUUGGUC	-2.1	-2.4	-3.3	1	0	2	-34.2	71.9	3.4	3	III	36.8	43.1	52.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1605	NM_006357	Huesken	UBE2E3	10477	UAUAGUUGAUCUCCAUUCAua	UGAAUGGAGAUCAACUAUA	-2.1	-1.3	0.1	1	1	2	-33.3	89.1	-0.7	8	II	31.6	65	90.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1606	NM_006357	Huesken	UBE2E3	10477	AUUCAUAAAUGUUAUCUCCuu	GGAGAUAACAUUUAUGAAU	-3.3	-1.1	1.6	2	4	2	-29.6	93.9	13.1	7	Ia	26.3	73.7	75.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1607	NM_006357	Huesken	UBE2E3	10477	UCAUAAAUGUUAUCUCCUUua	AAGGAGAUAACAUUUAUGA	-0.9	-2.4	1.6	1	1	2	-30.6	76.8	-1.3	7	II	26.3	60.8	52	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1608	NM_006357	Huesken	UBE2E3	10477	UAUCUCCUUUAGGCCCAGCac	GCUGGGCCUAAAGGAGAUA	-3.4	-1.3	-0.7	1	4	5	-40.8	68.3	9.7	6	Ib	52.6	65.9	67.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1609	NM_006357	Huesken	UBE2E3	10477	UCCUUUAGGCCCAGCACUGca	CAGUGCUGGGCCUAAAGGA	-2.1	-2.4	-0.7	0	2	5	-42.4	52.9	-7.3	5	Ia	57.9	59.5	73.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1610	NM_006357	Huesken	UBE2E3	10477	AGGCCCAGCACUGCAAUUAgg	UAAUUGCAGUGCUGGGCCU	-1.3	-2.1	0	3	-2	5	-41.2	53	-3.1	5	II	52.6	43.1	47.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1611	NM_006357	Huesken	UBE2E3	10477	CCAGCACUGCAAUUAGGAGga	CUCCUAAUUGCAGUGCUGG	-2.1	-3.3	0.8	0	1	2	-39	48.7	2.6	1	II	52.6	35	51.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1612	NM_006357	Huesken	UBE2E3	10477	GCACUGCAAUUAGGAGGAGga	CUCCUCCUAAUUGCAGUGC	-2.1	-3.4	2.1	0	1	2	-39.3	27.6	0	2	II	52.6	27	34.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1613	NM_006357	Huesken	UBE2E3	10477	CACUGCAAUUAGGAGGAGGau	CCUCCUCCUAAUUGCAGUG	-3.3	-2.1	2.1	-1	0	2	-39.2	52.7	0.3	3	II	52.6	46.8	62.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1614	NM_006357	Huesken	UBE2E3	10477	AUUAGGAGGAGGAUCAAGGgu	CCUUGAUCCUCCUCCUAAU	-3.3	-1.1	1.9	2	5	2	-38.6	66.8	-0.3	7	Ia	47.4	70.3	81.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1615	NM_006357	Huesken	UBE2E3	10477	AUCAAGGGUUAUUUCAGCUag	AGCUGAAAUAACCCUUGAU	-2.1	-1.1	3.4	1	3	3	-34.6	77.9	6.4	6	II	36.8	61	82	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1616	NM_006357	Huesken	UBE2E3	10477	GGUUAUUUCAGCUAGCUCCuu	GGAGCUAGCUGAAAUAACC	-3.3	-3.3	0.5	2	1	2	-37.3	80.7	7.8	6	II	47.4	58.8	80.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1617	NM_006357	Huesken	UBE2E3	10477	UAUUUCAGCUAGCUCCUUCug	GAAGGAGCUAGCUGAAAUA	-2.4	-1.3	0.5	1	4	2	-36.3	90.6	14.4	9	Ia	42.1	73.9	88.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1618	NM_006357	Huesken	UBE2E3	10477	UUCAGCUAGCUCCUUCUGAau	UCAGAAGGAGCUAGCUGAA	-2.4	-0.9	-1.7	2	1	2	-39.6	70.4	-1.4	6	II	47.4	60.3	87.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1619	NM_006357	Huesken	UBE2E3	10477	CAGCUAGCUCCUUCUGAAUuc	AUUCAGAAGGAGCUAGCUG	-1.1	-2.1	0.5	0	-3	2	-38.3	58.9	-5.3	3	III	47.4	36.7	42.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1620	NM_006357	Huesken	UBE2E3	10477	AGCUAGCUCCUUCUGAAUUcu	AAUUCAGAAGGAGCUAGCU	-0.9	-2.1	0.7	2	0	2	-37.1	69.4	6.5	4	II	42.1	46.1	71.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1621	NM_006357	Huesken	UBE2E3	10477	UUCUGAAUUCUUUUAGCACua	GUGCUAAAAGAAUUCAGAA	-2.2	-0.9	2.6	0	4	2	-31.1	89.6	12.8	6	Ia	31.6	70.2	81.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1622	NM_006357	Huesken	UBE2E3	10477	ACUAGUGGAUAACUUAGCAgu	UGCUAAGUUAUCCACUAGU	-2.1	-2.2	0.4	1	2	2	-35.1	68.2	3.3	5	II	36.8	49.3	82.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1623	NM_006357	Huesken	UBE2E3	10477	GCAGUGGUUUUGCUAGAGAgu	UCUCUAGCAAAACCACUGC	-2.4	-3.4	-0.3	0	-1	2	-38	46.5	1	3	III	47.4	33.6	35.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1624	NM_006357	Huesken	UBE2E3	10477	GUGGUUUUGCUAGAGAGUUug	AACUCUCUAGCAAAACCAC	-0.9	-2.2	2.6	2	-1	2	-35.6	57.1	-4	3	II	42.1	47	52.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1625	NM_006357	Huesken	UBE2E3	10477	GUUUUGCUAGAGAGUUUGGug	CCAAACUCUCUAGCAAAAC	-3.3	-2.2	2.6	-1	2	2	-34.3	69.7	-0.3	5	II	42.1	46.9	69.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1626	NM_006357	Huesken	UBE2E3	10477	UUUGCUAGAGAGUUUGGUGuu	CACCAAACUCUCUAGCAAA	-2.1	-0.9	2.6	0	5	2	-35.5	84.4	4.6	6	Ia	42.1	72.1	84.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1627	NM_006357	Huesken	UBE2E3	10477	UUGCUAGAGAGUUUGGUGUuu	ACACCAAACUCUCUAGCAA	-2.2	-0.9	2.6	2	2	2	-36.8	79.9	1.4	8	II	42.1	65.2	90.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1628	NM_006357	Huesken	UBE2E3	10477	CUGGGUGGCAGAAGGUUUUcu	AAAACCUUCUGCCACCCAG	-0.9	-2.1	0.6	-1	-3	3	-39.6	45.8	-8.6	4	III	52.6	42.6	32.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1629	NM_006357	Huesken	UBE2E3	10477	UCUUUCCUCUUGUUCUUCAgg	UGAAGAACAAGAGGAAAGA	-2.1	-2.4	2.1	1	1	2	-34.6	88.4	-1.4	7	II	36.8	57.4	89.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1630	NM_006357	Huesken	UBE2E3	10477	UCCUCUUGUUCUUCAGGCUcu	AGCCUGAAGAACAAGAGGA	-2.1	-2.4	-0.9	1	1	3	-39.2	75.8	-3.3	5	II	47.4	56.4	89	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1631	NM_006357	Huesken	UBE2E3	10477	UGGAGCGGCUGGGUCUCGCug	GCGAGACCCAGCCGCUCCA	-3.4	-2.1	-2.1	1	4	5	-48.1	43.6	9.7	3	II	73.7	57.1	69.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1632	NM_006357	Huesken	UBE2E3	10477	CGGCUGGGUCUCGCUGGUCcg	GACCAGCGAGACCCAGCCG	-2.4	-2.4	-1	0	-1	4	-46.9	25	3.1	0	II	73.7	37.8	54.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1633	NM_006357	Huesken	UBE2E3	10477	CGCAUCUGAACUGCCACUGcu	CAGUGGCAGUUCAGAUGCG	-2.1	-2.4	-2.4	0	-1	3	-40.5	57.5	-1.3	2	II	57.9	47.2	53	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1634	NM_021988	Huesken	UBE2V1	7335	GUCCUUCGGGCGGCUGAGGga	CCUCAGCCGCCCGAAGGAC	-3.3	-2.2	-0.9	2	1	8	-46.8	32.7	-2.4	1	II	73.7	37.3	55.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1635	NM_021988	Huesken	UBE2V1	7335	CCUUCGGGCGGCUGAGGGAgu	UCCCUCAGCCGCCCGAAGG	-2.4	-3.3	0.2	-1	-2	8	-47.9	27.8	-8.1	2	III	73.7	24	46.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1636	NM_021988	Huesken	UBE2V1	7335	UUUCAUAUUUUCUUUAGACau	GUCUAAAGAAAAUAUGAAA	-2.2	-0.9	1	1	5	1	-26.8	101.1	10.1	7	Ia	21.1	77.5	75.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1637	NM_021988	Huesken	UBE2V1	7335	UUCUUUAGACAUCAUUAGGcg	CCUAAUGAUGUCUAAAGAA	-3.3	-0.9	1	0	3	2	-31.6	95.8	7.7	7	Ia	31.6	74.1	95.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1638	NM_021988	Huesken	UBE2V1	7335	UCUUUAGACAUCAUUAGGCgc	GCCUAAUGAUGUCUAAAGA	-3.4	-2.4	1.8	1	5	3	-34.1	81.4	4.8	8	Ia	36.8	73.4	94.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1639	NM_021988	Huesken	UBE2V1	7335	AGACAUCAUUAGGCGCCGAag	UCGGCGCCUAAUGAUGUCU	-2.4	-2.1	-1.4	2	1	7	-40.4	43.2	1.3	3	II	52.6	40.6	78.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1640	NM_021988	Huesken	UBE2V1	7335	UUAGGCGCCGAAGCUCUUGca	CAAGAGCUUCGGCGCCUAA	-2.1	-0.9	-4.5	1	3	7	-40.9	54.3	2.3	5	II	57.9	68.1	84.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1641	NM_021988	Huesken	UBE2V1	7335	UAGGCGCCGAAGCUCUUGCag	GCAAGAGCUUCGGCGCCUA	-3.4	-1.3	-4.5	1	4	7	-43.4	63.4	14.4	4	II	63.2	67.3	82.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1642	NM_021988	Huesken	UBE2V1	7335	CGAAGCUCUUGCAGGACAAcu	UUGUCCUGCAAGAGCUUCG	-0.9	-2.4	-0.5	-1	-3	2	-39.3	43.7	-10.7	3	III	52.6	30.2	51.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1643	NM_021988	Huesken	UBE2V1	7335	CUCUUGCAGGACAACUUUGau	CAAAGUUGUCCUGCAAGAG	-2.1	-2.1	-0.1	0	-1	2	-36.3	60.8	-4.6	4	II	47.4	50.3	71.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1644	NM_021988	Huesken	UBE2V1	7335	UCUUGCAGGACAACUUUGAug	UCAAAGUUGUCCUGCAAGA	-2.4	-2.4	-0.1	1	0	2	-36.6	66.9	-8.7	8	II	42.1	55.6	89.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1645	NM_021988	Huesken	UBE2V1	7335	GGACAACUUUGAUGCUAUAug	UAUAGCAUCAAAGUUGUCC	-1.3	-3.3	-1.4	0	-3	2	-33.9	50.3	-1.1	1	III	36.8	24	47	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1646	NM_021988	Huesken	UBE2V1	7335	ACAACUUUGAUGCUAUAUGaa	CAUAUAGCAUCAAAGUUGU	-2.1	-2.2	-1.3	3	3	2	-31.4	85.8	4.7	5	Ia	31.6	66.1	77.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1647	NM_021988	Huesken	UBE2V1	7335	CAACUUUGAUGCUAUAUGAau	UCAUAUAGCAUCAAAGUUG	-2.4	-2.1	0	1	-2	2	-31.6	78.4	-8.1	5	II	31.6	49.3	49	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1648	NM_021988	Huesken	UBE2V1	7335	CUUUGAUGCUAUAUGAAUUcu	AAUUCAUAUAGCAUCAAAG	-0.9	-2.1	3.2	1	-1	2	-29.3	85.6	-0.9	4	II	26.3	50.8	70	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1649	NM_021988	Huesken	UBE2V1	7335	UGAUGCUAUAUGAAUUCUGcc	CAGAAUUCAUAUAGCAUCA	-2.1	-2.1	2.1	1	3	2	-32	80.2	-2.6	7	Ib	31.6	73.2	90	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1650	NM_021988	Huesken	UBE2V1	7335	GAUGCUAUAUGAAUUCUGCca	GCAGAAUUCAUAUAGCAUC	-3.4	-2.4	0.1	0	3	2	-33.3	80.1	7.1	4	II	36.8	56.6	82	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1651	NM_021988	Huesken	UBE2V1	7335	UGCUAUAUGAAUUCUGCCAuu	UGGCAGAAUUCAUAUAGCA	-2.1	-2.1	0.1	1	0	3	-35.2	87.5	4.3	7	II	36.8	62.4	89.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1652	NM_021988	Huesken	UBE2V1	7335	GAAUUCUGCCAUUUUGCUAgc	UAGCAAAAUGGCAGAAUUC	-1.3	-2.4	-1.3	2	0	3	-33.4	83.9	3.9	4	II	36.8	44.5	63	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1653	NM_021988	Huesken	UBE2V1	7335	UUCUGCCAUUUUGCUAGCAcu	UGCUAGCAAAAUGGCAGAA	-2.1	-0.9	-4.6	0	1	3	-36.6	67.7	-0.7	4	II	42.1	54.8	71.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1654	NM_021988	Huesken	UBE2V1	7335	UGCCAUUUUGCUAGCACUGau	CAGUGCUAGCAAAAUGGCA	-2.1	-2.1	-4	1	2	3	-37.6	65.4	0.8	4	Ib	47.4	64.1	87.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1655	NM_021988	Huesken	UBE2V1	7335	UUGCUAGCACUGAUAUGGCuc	GCCAUAUCAGUGCUAGCAA	-3.4	-0.9	0.4	1	5	3	-38.5	64	7.1	6	Ib	47.4	69.1	92.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1656	NM_021988	Huesken	UBE2V1	7335	GCUAGCACUGAUAUGGCUCuu	GAGCCAUAUCAGUGCUAGC	-2.4	-3.4	-0.5	0	2	3	-40	61.1	12.5	3	II	52.6	47.9	90.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1657	NM_021988	Huesken	UBE2V1	7335	CUAGCACUGAUAUGGCUCUug	AGAGCCAUAUCAGUGCUAG	-2.1	-2.1	-0.5	-1	-1	3	-38.7	54.5	-6	4	III	47.4	46.8	90.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1658	NM_021988	Huesken	UBE2V1	7335	UAGCACUGAUAUGGCUCUUgg	AAGAGCCAUAUCAGUGCUA	-0.9	-1.3	-0.5	1	1	3	-37.5	65.2	4.1	6	II	42.1	55.2	84.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1659	NM_021988	Huesken	UBE2V1	7335	UAUGGCUCUUGGGUCCACCac	GGUGGACCCAAGAGCCAUA	-3.3	-1.3	-2.3	1	5	3	-42.9	65.9	9.7	4	Ib	57.9	72.2	84.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1660	NM_021988	Huesken	UBE2V1	7335	UUGGGUCCACCACUCCAUUag	AAUGGAGUGGUGGACCCAA	-0.9	-0.9	-1.3	0	1	3	-41.3	57.4	-1.6	3	II	52.6	57.6	83.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1661	NM_021988	Huesken	UBE2V1	7335	CCACCACUCCAUUAGAACUau	AGUUCUAAUGGAGUGGUGG	-2.1	-3.3	0.3	1	-2	2	-38.1	60.1	-0.7	1	III	47.4	40.4	56.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1662	NM_021988	Huesken	UBE2V1	7335	GUUACAAAUCUUACAAAGGgg	CCUUUGUAAGAUUUGUAAC	-3.3	-2.2	1.8	0	2	2	-29.8	74.8	0.4	4	II	31.6	57.3	63.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1663	NM_021988	Huesken	UBE2V1	7335	GGUCCACAUUCUAUUUUAAgg	UUAAAAUAGAAUGUGGACC	-0.9	-3.3	3.2	1	-2	2	-31.4	67.9	-3.3	2	III	31.6	23.7	38.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1664	NM_021988	Huesken	UBE2V1	7335	UAUUUUAAGGCUGUAUAUUcg	AAUAUACAGCCUUAAAAUA	-0.9	-1.3	2.6	1	2	3	-28.2	96.4	-1	8	II	21.1	68.7	82.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1665	NM_021988	Huesken	UBE2V1	7335	UUUUAAGGCUGUAUAUUCGgu	CGAAUAUACAGCCUUAAAA	-2.4	-0.9	2.6	1	5	3	-30.6	90.9	5.5	9	Ia	31.6	74.7	89.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1666	NM_021988	Huesken	UBE2V1	7335	UAAGGCUGUAUAUUCGGUUuu	AACCGAAUAUACAGCCUUA	-0.9	-1.3	3.4	1	3	3	-34.3	76.4	-1.6	7	II	36.8	70.6	96.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1667	NM_021988	Huesken	UBE2V1	7335	CAUAAAUUGUUCUUGGAGGcc	CCUCCAAGAACAAUUUAUG	-3.3	-2.1	2.5	1	2	2	-32.1	79	0.7	5	II	36.8	57.7	63.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1668	NM_021988	Huesken	UBE2V1	7335	AUAAAUUGUUCUUGGAGGCcc	GCCUCCAAGAACAAUUUAU	-3.4	-1.1	2.5	1	5	3	-33.4	79.7	10.8	7	Ia	36.8	68.8	76.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1669	NM_021988	Huesken	UBE2V1	7335	AAUUGUUCUUGGAGGCCCAau	UGGGCCUCCAAGAACAAUU	-2.1	-0.9	-1.3	1	2	5	-38.8	62.7	-6.3	5	II	47.4	53	57.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1670	NM_021988	Huesken	UBE2V1	7335	GGAGGCCCAAUUAUCAUCCcu	GGAUGAUAAUUGGGCCUCC	-3.3	-3.3	0.8	0	2	5	-39.8	58.6	10.2	3	II	52.6	52.7	70.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1671	NM_021988	Huesken	UBE2V1	7335	UUAUCAUCCCUGUCCAUCUug	AGAUGGACAGGGAUGAUAA	-2.1	-0.9	0.8	2	2	3	-37.7	84.5	-4	8	II	42.1	73.6	90	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1672	NM_021988	Huesken	UBE2V1	7335	CAUCCCUGUCCAUCUUGUAag	UACAAGAUGGACAGGGAUG	-1.3	-2.1	1.9	1	-2	3	-38.5	61.6	-8.1	3	III	47.4	37.3	52.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1673	NM_021988	Huesken	UBE2V1	7335	UCAUGUCUUCGUCAUCUUCua	GAAGAUGACGAAGACAUGA	-2.4	-2.4	1	0	2	2	-35.7	79	12.8	6	Ib	42.1	64.3	95	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1674	NM_021988	Huesken	UBE2V1	7335	UGUCUUCGUCAUCUUCUAGac	CUAGAAGAUGACGAAGACA	-2.1	-2.1	2.8	1	3	2	-35.6	83.5	10.1	6	Ib	42.1	59	88.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1675	NM_021988	Huesken	UBE2V1	7335	GUCUUCGUCAUCUUCUAGAcc	UCUAGAAGAUGACGAAGAC	-2.4	-2.2	1.2	1	-1	2	-35.9	77.1	-3.8	5	III	42.1	46	58	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1676	NM_021988	Huesken	UBE2V1	7335	UCUUCGUCAUCUUCUAGACcc	GUCUAGAAGAUGACGAAGA	-2.2	-2.4	-2.7	1	3	2	-35.9	85.2	8.1	5	Ib	42.1	61.1	92	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1677	NM_021988	Huesken	UBE2V1	7335	UUCGUCAUCUUCUAGACCCca	GGGUCUAGAAGAUGACGAA	-3.3	-0.9	-2.7	1	4	3	-38	77.6	5.1	5	Ib	47.4	79.5	96.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1678	NM_021988	Huesken	UBE2V1	7335	CGUCAUCUUCUAGACCCCAgc	UGGGGUCUAGAAGAUGACG	-2.1	-2.4	-2.7	0	-2	4	-40.1	40.3	-3.1	1	III	52.6	37.9	38.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1679	NM_021988	Huesken	UBE2V1	7335	CAUCUUCUAGACCCCAGCUaa	AGCUGGGGUCUAGAAGAUG	-2.1	-2.1	-0.8	0	0	4	-40.7	62.6	-3.6	3	II	52.6	47.7	73	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1680	NM_021988	Huesken	UBE2V1	7335	AUCUUCUAGACCCCAGCUAac	UAGCUGGGGUCUAGAAGAU	-1.3	-1.1	-0.8	3	0	4	-39.9	66	-6.1	7	II	47.4	51.4	82.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1681	NM_021988	Huesken	UBE2V1	7335	GCUAACUGUGCCAUCUCCUac	AGGAGAUGGCACAGUUAGC	-2.1	-3.4	1	1	1	3	-40.6	55.9	-6.2	4	II	52.6	44.3	60.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1682	NM_021988	Huesken	UBE2V1	7335	UAACUGUGCCAUCUCCUACuc	GUAGGAGAUGGCACAGUUA	-2.2	-1.3	0.8	3	4	3	-38.6	81.3	17.5	8	Ia	47.4	72.9	88.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1683	NM_021988	Huesken	UBE2V1	7335	UGUGCCAUCUCCUACUCCUuu	AGGAGUAGGAGAUGGCACA	-2.1	-2.1	0.6	1	2	3	-42	68	-8.7	5	II	52.6	61.4	92.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1684	NM_021988	Huesken	UBE2V1	7335	CCAUCUCCUACUCCUUUCUgg	AGAAAGGAGUAGGAGAUGG	-2.1	-3.3	3	-1	-2	2	-38.5	71.5	-2.9	4	III	47.4	43.6	58.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1685	NM_021988	Huesken	UBE2V1	7335	UACUCCUUUCUGGCCUUCUuc	AGAAGGCCAGAAAGGAGUA	-2.1	-1.3	-0.5	1	1	4	-39.3	78.4	-4	7	II	47.4	64.9	86.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1686	NM_021988	Huesken	UBE2V1	7335	GGCCUUCUUCGAGUUCUUCca	GAAGAACUCGAAGAAGGCC	-2.4	-3.3	1	1	0	4	-38.6	57.8	9.7	2	II	52.6	38.3	53.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1687	NM_021988	Huesken	UBE2V1	7335	CCUUCUUCGAGUUCUUCCAac	UGGAAGAACUCGAAGAAGG	-2.1	-3.3	1	1	-2	2	-37.3	77.5	-2.7	4	II	47.4	43.7	59.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1688	NM_021988	Huesken	UBE2V1	7335	GUUCUUCCAACAGUCGGAAau	UUCCGACUGUUGGAAGAAC	-0.9	-2.2	-3.4	1	0	3	-37.2	55.8	-3.8	3	II	47.4	34.3	52.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1689	NM_021988	Huesken	UBE2V1	7335	UUCCAACAGUCGGAAAUUGcg	CAAUUUCCGACUGUUGGAA	-2.1	-0.9	-2.8	2	2	3	-34.6	56.1	0.4	5	Ib	42.1	58.5	95.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1690	NM_021988	Huesken	UBE2V1	7335	CAACAGUCGGAAAUUGCGAgg	UCGCAAUUUCCGACUGUUG	-2.4	-2.1	0.2	2	0	3	-36.2	51.5	-1	4	II	47.4	46.5	85.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1691	NM_021988	Huesken	UBE2V1	7335	CAGUCGGAAAUUGCGAGGGac	CCCUCGCAAUUUCCGACUG	-3.3	-2.1	0.3	-1	1	3	-39.7	52.5	1	3	II	57.9	51.3	84.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1692	NM_021988	Huesken	UBE2V1	7335	UCGGAAAUUGCGAGGGACUuu	AGUCCCUCGCAAUUUCCGA	-2.1	-2.4	0.8	-1	0	3	-40	44.5	-6.2	5	II	52.6	54.2	96	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1693	NM_021988	Huesken	UBE2V1	7335	AUUGCGAGGGACUUUUACUcc	AGUAAAAGUCCCUCGCAAU	-2.1	-1.1	1.1	2	3	3	-36	74.9	-1.7	5	II	42.1	61.7	91	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1694	NM_021988	Huesken	UBE2V1	7335	UGCGAGGGACUUUUACUCCuu	GGAGUAAAAGUCCCUCGCA	-3.3	-2.1	-0.8	0	3	3	-39.7	63.4	15.5	6	II	52.6	65.9	88.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1695	NM_021988	Huesken	UBE2V1	7335	GAGGGACUUUUACUCCUUUgc	AAAGGAGUAAAAGUCCCUC	-0.9	-2.4	-1	0	-1	3	-35.7	54.5	0.8	1	III	42.1	37.6	65.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1696	NM_021988	Huesken	UBE2V1	7335	UUUACUCCUUUGCAGUGAAau	UUCACUGCAAAGGAGUAAA	-0.9	-0.9	0	0	1	2	-34.3	76.3	-1.4	6	II	36.8	53.9	81.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1697	NM_021988	Huesken	UBE2V1	7335	UACUCCUUUGCAGUGAAAUuu	AUUUCACUGCAAAGGAGUA	-1.1	-1.3	-0.9	1	1	2	-34.5	82.9	-4	6	II	36.8	59.9	95	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1698	NM_021988	Huesken	UBE2V1	7335	CUCCUUUGCAGUGAAAUUUuc	AAAUUUCACUGCAAAGGAG	-0.9	-2.1	0.3	2	-3	2	-32.8	74.7	-3	5	II	36.8	51.1	45.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1699	NM_021988	Huesken	UBE2V1	7335	CCUUUGCAGUGAAAUUUUCua	GAAAAUUUCACUGCAAAGG	-2.4	-3.3	0.3	0	-1	2	-31.6	65.8	2.8	4	II	36.8	44.4	83.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1700	NM_021988	Huesken	UBE2V1	7335	UUGCAGUGAAAUUUUCUAGgu	CUAGAAAAUUUCACUGCAA	-2.1	-0.9	1	2	3	2	-30.8	94.3	10.4	8	Ib	31.6	73.5	96	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1701	NM_021988	Huesken	UBE2V1	7335	GCAGUGAAAUUUUCUAGGUuc	ACCUAGAAAAUUUCACUGC	-2.2	-3.4	1	0	0	2	-33.3	62.6	-1	3	III	36.8	41.8	36.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1702	NM_021988	Huesken	UBE2V1	7335	CAGUGAAAUUUUCUAGGUUca	AACCUAGAAAAUUUCACUG	-0.9	-2.1	1	-1	-1	2	-30.8	68.4	1.8	3	II	31.6	43.5	49.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1703	NM_021988	Huesken	UBE2V1	7335	GUGAAAUUUUCUAGGUUCAag	UGAACCUAGAAAAUUUCAC	-2.1	-2.2	0.9	0	-2	2	-31.1	70.7	-3	5	II	31.6	42.7	43	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1704	NM_021988	Huesken	UBE2V1	7335	GAAAUUUUCUAGGUUCAAGuc	CUUGAACCUAGAAAAUUUC	-2.1	-2.4	1.9	2	2	2	-29.8	85.9	4.6	5	II	31.6	54.6	53.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1705	NM_021988	Huesken	UBE2V1	7335	AUUUUCUAGGUUCAAGUCUuc	AGACUUGAACCUAGAAAAU	-2.1	-1.1	1.4	3	2	2	-32.3	86.6	1.5	8	II	31.6	66	71.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1706	NM_021988	Huesken	UBE2V1	7335	UCUAGGUUCAAGUCUUCCUuc	AGGAAGACUUGAACCUAGA	-2.1	-2.4	-0.2	2	2	2	-37.2	82.8	-1.7	7	II	42.1	67.6	78	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1707	NM_021988	Huesken	UBE2V1	7335	AGUCUUCCUUCAUCACUCAgu	UGAGUGAUGAAGGAAGACU	-2.1	-2.1	1.3	2	0	2	-37.3	65.5	-6.1	5	II	42.1	46.8	62.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1708	NM_022005	Huesken	Fxyd6	53826	GCUUGGUUCGUUUCCAAGGga	CCUUGGAAACGAACCAAGC	-3.3	-3.4	-4.3	2	1	2	-37.8	64.9	3	2	II	52.6	50.2	85.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1709	NM_022005	Huesken	Fxyd6	53826	CAAGGGAACACAAUGGGCAgg	UGCCCAUUGUGUUCCCUUG	-2.1	-2.1	1.9	0	0	4	-39.8	44.3	-8.4	4	III	52.6	49.7	95.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1710	NM_022005	Huesken	Fxyd6	53826	GGUGGGGCAGGCUCUGGUGug	CACCAGAGCCUGCCCCACC	-2.1	-3.3	0.8	0	1	5	-48.1	32.3	-2.4	1	II	73.7	34.5	90.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1711	NM_022005	Huesken	Fxyd6	53826	GUGGGGCAGGCUCUGGUGUga	ACACCAGAGCCUGCCCCAC	-2.2	-2.2	0.8	1	0	5	-47	36.2	1.4	1	III	68.4	34.2	72.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1712	NM_022005	Huesken	Fxyd6	53826	GUGUGAGUUUCCACUCUGGug	CCAGAGUGGAAACUCACAC	-3.3	-2.2	-5.9	0	1	2	-39.1	54	-7.3	3	II	52.6	46.7	95.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1713	NM_022005	Huesken	Fxyd6	53826	GUGAGUUUCCACUCUGGUGac	CACCAGAGUGGAAACUCAC	-2.1	-2.2	-5.9	2	1	2	-39.1	68.4	-0.1	2	II	52.6	51.9	92.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1714	NM_022005	Huesken	Fxyd6	53826	GUUUCCACUCUGGUGACCUuu	AGGUCACCAGAGUGGAAAC	-2.1	-2.2	-1.7	0	2	2	-40.3	62.2	-1.6	3	II	52.6	48.2	94.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1715	NM_022005	Huesken	Fxyd6	53826	CUCUGGUGACCUUUGACUCcc	GAGUCAAAGGUCACCAGAG	-2.4	-2.1	0.5	0	1	2	-39.3	67.9	8.4	3	II	52.6	52.8	85.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1716	NM_022005	Huesken	Fxyd6	53826	ACCUUUGACUCCCGUAGGCgu	GCCUACGGGAGUCAAAGGU	-3.4	-2.2	-3.3	1	3	4	-41.9	53.6	5.1	3	Ib	57.9	49.7	91.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1717	NM_022005	Huesken	Fxyd6	53826	UUGACUCCCGUAGGCGUUUgc	AAACGCCUACGGGAGUCAA	-0.9	-0.9	-3.3	0	0	4	-39.8	56.5	-4	5	II	52.6	57.6	93.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1718	NM_022005	Huesken	Fxyd6	53826	CACGGGUGUGCAUACCCUGuu	CAGGGUAUGCACACCCGUG	-2.1	-2.1	-5.6	0	0	4	-42.8	39.1	-4.4	0	II	63.2	49.9	70.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1719	NM_022005	Huesken	Fxyd6	53826	GGUGUGCAUACCCUGUUGCuc	GCAACAGGGUAUGCACACC	-3.4	-3.3	-1.9	0	2	3	-41.4	56.6	7.4	2	II	57.9	44.8	72.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1720	NM_022005	Huesken	Fxyd6	53826	CAUACCCUGUUGCUCUCUCcc	GAGAGAGCAACAGGGUAUG	-2.4	-2.1	1	0	2	3	-39.6	66.4	10.1	3	II	52.6	48.9	85.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1721	NM_022005	Huesken	Fxyd6	53826	CCUGUUGCUCUCUCCCGCUgc	AGCGGGAGAGAGCAACAGG	-2.1	-3.3	-1.9	-1	-1	5	-44.1	46.1	-8.3	1	III	63.2	43.8	43.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1722	NM_022005	Huesken	Fxyd6	53826	GUUGCUCUCUCCCGCUGCCcc	GGCAGCGGGAGAGAGCAAC	-3.3	-2.2	-1.9	0	3	5	-45.4	47.3	5.1	0	II	68.4	44	61.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1723	NM_022005	Huesken	Fxyd6	53826	CUCUCUCCCGCUGCCCCUGaa	CAGGGGCAGCGGGAGAGAG	-2.1	-2.1		0	0	5	-47.6	45.3	-1.3	1	II	73.7	43.9	66.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1724	NM_022005	Huesken	Fxyd6	53826	CACUCCGACCCCUGACAGGca	CCUGUCAGGGGUCGGAGUG	-3.3	-2.1	-1.9	-1	0	4	-45.3	43	-6.7	1	II	68.4	40.7	72.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1725	NM_022005	Huesken	Fxyd6	53826	ACCCCUGACAGGCAGGGGAaa	UCCCCUGCCUGUCAGGGGU	-2.4	-2.2	-9.7	3	0	4	-48.3	38.3	-1.4	2	II	68.4	38.7	42.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1726	NM_022005	Huesken	Fxyd6	53826	GGGAAAGAAACUGUAGUCUgc	AGACUACAGUUUCUUUCCC	-2.1	-3.3	0.2	0	-1	3	-35.8	57.6	1.7	5	III	42.1	39.9	55.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1727	NM_022005	Huesken	Fxyd6	53826	AAAGAAACUGUAGUCUGCCag	GGCAGACUACAGUUUCUUU	-3.3	-0.9	-0.1	1	6	3	-35.6	60.8	12.4	6	Ia	42.1	71.2	86	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1728	NM_022005	Huesken	Fxyd6	53826	AACUGUAGUCUGCCAGAUUcc	AAUCUGGCAGACUACAGUU	-0.9	-0.9	-1.7	1	0	3	-37	63.4	-1.6	4	II	42.1	52.3	81.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1729	NM_022005	Huesken	Fxyd6	53826	CAGAUUCCACACCAUGUGGca	CCACAUGGUGUGGAAUCUG	-3.3	-2.1	-4.9	-1	0	2	-39.1	64.6	0.4	3	II	52.6	50.9	40.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1730	NM_022005	Huesken	Fxyd6	53826	CACACCAUGUGGCAGAGAGac	CUCUCUGCCACAUGGUGUG	-2.1	-2.1	-2.4	0	0	3	-41.4	55.5	0.4	1	II	57.9	42	46.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1731	NM_022005	Huesken	Fxyd6	53826	AUGUGGCAGAGAGACAGAGcc	CUCUGUCUCUCUGCCACAU	-2.1	-1.1	1.8	2	3	3	-40.7	47.7	0	5	II	52.6	54.2	62.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1732	NM_022005	Huesken	Fxyd6	53826	GUGGCAGAGAGACAGAGCCac	GGCUCUGUCUCUCUGCCAC	-3.3	-2.2	-3.8	1	2	3	-44.2	41.5	9.8	1	II	63.2	50.6	45.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1733	NM_022005	Huesken	Fxyd6	53826	UGGCAGAGAGACAGAGCCAcc	UGGCUCUGUCUCUCUGCCA	-2.1	-2.1	-3.8	0	0	3	-44.1	35.9	-3.7	3	II	57.9	49.8	76.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1734	NM_022005	Huesken	Fxyd6	53826	GCAGAGAGACAGAGCCACCgg	GGUGGCUCUGUCUCUCUGC	-3.3	-3.4	0.5	-1	1	3	-44.2	30.2	9.8	2	II	63.2	44.3	71.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1735	NM_022005	Huesken	Fxyd6	53826	CCUGCUGCCUGUCCCCUGGgc	CCAGGGGACAGGCAGCAGG	-3.3	-3.3	0.4	0	0	4	-48	46.5	3.4	2	II	73.7	42.3	45.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1736	NM_022005	Huesken	Fxyd6	53826	CCCUGGGCUAUCUCCUGCCac	GGCAGGAGAUAGCCCAGGG	-3.3	-3.3	-3.4	-1	0	4	-46.4	44.1	3.1	1	II	68.4	48.3	67.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1737	NM_022005	Huesken	Fxyd6	53826	CUGCCACAGACCAGCUUCAga	UGAAGCUGGUCUGUGGCAG	-2.1	-2.1	-2.2	0	-3	3	-42.4	38.7	-13	1	III	57.9	35.7	44.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1738	NM_022005	Huesken	Fxyd6	53826	CUUCAGAGAAGGAAGGAGUcc	ACUCCUUCCUUCUCUGAAG	-2.2	-2.1	-0.6	0	-1	2	-38.2	52.5	-5.9	5	II	47.4	48.9	50.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1739	NM_022005	Huesken	Fxyd6	53826	AGAGAAGGAAGGAGUCCUCug	GAGGACUCCUUCCUUCUCU	-2.4	-2.1	-3.3	0	3	2	-40.9	42.7	7.5	5	Ib	52.6	52.6	81.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1740	NM_022005	Huesken	Fxyd6	53826	GCUAGAGAUGGAAGUCUCAga	UGAGACUUCCAUCUCUAGC	-2.1	-3.4	-2.8	-1	-2	2	-38.9	40.8	-6.3	3	III	47.4	31.3	32.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1741	NM_022005	Huesken	Fxyd6	53826	GAAGUCUCAGAAGCAGGGAag	UCCCUGCUUCUGAGACUUC	-2.4	-2.4	1.7	1	0	3	-40.7	46.6	-1.5	4	II	52.6	38.7	67.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1742	NM_022005	Huesken	Fxyd6	53826	AGGGAAGGUGGAUGAAGGCag	GCCUUCAUCCACCUUCCCU	-3.4	-2.1		0	3	3	-42.7	36.7	10.1	3	II	57.9	48.7	85.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1743	NM_022005	Huesken	Fxyd6	53826	GCAGAAGGGCUCCAAAGCCua	GGCUUUGGAGCCCUUCUGC	-3.3	-3.4	-4.5	1	2	4	-43.5	27.1	10.2	1	II	63.2	43.6	73.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1744	NM_022005	Huesken	Fxyd6	53826	GCUCCAAAGCCUAUGGGGAua	UCCCCAUAGGCUUUGGAGC	-2.4	-3.4	-2	2	0	4	-42.8	41.2	-3.3	0	III	57.9	27.5	35.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1745	NM_022005	Huesken	Fxyd6	53826	UGGGGAUAUUGCUCAGAAGuu	CUUCUGAGCAAUAUCCCCA	-2.1	-2.1	1.1	1	1	4	-38.4	60.5	-2.4	4	II	47.4	55.1	93	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1746	NM_022005	Huesken	Fxyd6	53826	UAUUGCUCAGAAGUUAAGUaa	ACUUAACUUCUGAGCAAUA	-2.2	-1.3	-0.2	1	3	2	-32.5	87.5	0.7	7	II	31.6	67.8	90.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1747	NM_022005	Huesken	Fxyd6	53826	GUUAAGUAAGGAGCUCAUUcc	AAUGAGCUCCUUACUUAAC	-0.9	-2.2	-0.2	1	-1	2	-33.7	62.9	-1.3	5	II	36.8	45.2	70.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1748	NM_022005	Huesken	Fxyd6	53826	CUCAUUCCUCAGAGGCAAGaa	CUUGCCUCUGAGGAAUGAG	-2.1	-2.1	-1.2	-2	-1	3	-39.3	55.3	-2.3	2	II	52.6	42.5	59.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1749	NM_022005	Huesken	Fxyd6	53826	AAGAAUUCCUGGGUGUUCUgu	AGAACACCCAGGAAUUCUU	-2.1	-0.9	1	3	2	3	-36.7	77.9	1.1	7	II	42.1	60.9	84.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1750	NM_022005	Huesken	Fxyd6	53826	UGGGUGUUCUGUAAAGGGUgg	ACCCUUUACAGAACACCCA	-2.2	-2.1	2	2	1	3	-38.9	63.6	-3.5	5	II	47.4	58.1	86.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1751	NM_022005	Huesken	Fxyd6	53826	UAAAGGGUGGGGCGGAUUGgg	CAAUCCGCCCCACCCUUUA	-2.1	-1.3	4.3	0	3	7	-41.6	54.1	2.4	7	Ib	57.9	64.6	80.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1752	NM_022005	Huesken	Fxyd6	53826	CCUAAGUGGGCACCCUAAAag	UUUAGGGUGCCCACUUAGG	-0.9	-3.3	-2.5	1	-4	4	-40.1	43.8	-3	3	II	52.6	31.1	48.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1753	NM_022005	Huesken	Fxyd6	53826	AGACGCCUCAAGGGAGGCAau	UGCCUCCCUUGAGGCGUCU	-2.1	-2.1	-8.1	2	1	4	-45.3	33	-1.5	2	II	63.2	37.5	34.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1754	NM_022005	Huesken	Fxyd6	53826	GACGCCUCAAGGGAGGCAAuc	UUGCCUCCCUUGAGGCGUC	-0.9	-2.4	-8.1	1	-2	4	-44.1	42.1	-3.8	1	III	63.2	29.2	37.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1755	NM_022005	Huesken	Fxyd6	53826	CUGAAUCUCCAACUGCAAAcg	UUUGCAGUUGGAGAUUCAG	-0.9	-2.1	1.2	0	-3	2	-35.5	65.1	4	3	II	42.1	38.5	48.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1756	NM_022005	Huesken	Fxyd6	53826	CUUCUAAACUGGUGGAAUCug	GAUUCCACCAGUUUAGAAG	-2.4	-2.1	1.6	0	0	2	-34.7	65.8	5.4	4	II	42.1	50.4	84	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1757	NM_022005	Huesken	Fxyd6	53826	AAUCUGGAGCAAGUUUCUUcu	AAGAAACUUGCUCCAGAUU	-0.9	-0.9	1.6	3	3	2	-34.3	65.8	0.7	6	II	36.8	53.7	79	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1758	NM_022005	Huesken	Fxyd6	53826	UCUGGAGCAAGUUUCUUCUca	AGAAGAAACUUGCUCCAGA	-2.1	-2.4	1.6	0	2	2	-36.8	78.6	1.4	7	II	42.1	62.2	90	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1759	NM_022005	Huesken	Fxyd6	53826	UUUCUUCUCAGAGUCCACAag	UGUGGACUCUGAGAAGAAA	-2.1	-0.9	-0.7	2	2	2	-37.1	77.9	-1.4	7	II	42.1	66.9	85.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1760	NM_022005	Huesken	Fxyd6	53826	UCAGAGUCCACAAGGAACAuc	UGUUCCUUGUGGACUCUGA	-2.1	-2.4	-2.3	1	0	2	-39.3	51.5	-6	6	II	47.4	60.7	94.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1761	NM_022005	Huesken	Fxyd6	53826	GUCCACAAGGAACAUCAGUga	ACUGAUGUUCCUUGUGGAC	-2.2	-2.2	-2.3	2	-1	2	-38.1	49.9	3.4	2	III	47.4	41.2	55.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1762	NM_022005	Huesken	Fxyd6	53826	ACGAACUUUCUCAGCCCCGgg	CGGGGCUGAGAAAGUUCGU	-2.4	-2.2	-4.1	1	3	6	-40.8	49.1	-2.2	3	Ib	57.9	59.8	95.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1763	NM_022005	Huesken	Fxyd6	53826	GAACUUUCUCAGCCCCGGGau	CCCGGGGCUGAGAAAGUUC	-3.3	-2.4	-2.4	2	3	8	-42.8	54.2	4.7	2	II	63.2	52	85.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1764	NM_022005	Huesken	Fxyd6	53826	UCUCAGCCCCGGGAUCACUgg	AGUGAUCCCGGGGCUGAGA	-2.1	-2.4	0.9	2	1	8	-45.7	41.4	-1.3	4	II	63.2	50.2	94.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1765	NM_022005	Huesken	Fxyd6	53826	GGUACAACUCUCCCUCAAGcc	CUUGAGGGAGAGUUGUACC	-2.1	-3.3	0.6	0	0	3	-39.4	55.2	0.1	3	II	52.6	39.2	63.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1766	NM_022005	Huesken	Fxyd6	53826	CAACUCUCCCUCAAGCCUAaa	UAGGCUUGAGGGAGAGUUG	-1.3	-2.1	-2.3	2	-2	3	-40.5	49.9	-10.7	3	II	52.6	46.3	76.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1767	NM_022005	Huesken	Fxyd6	53826	UCUCCCUCAAGCCUAAAUCca	GAUUUAGGCUUGAGGGAGA	-2.4	-2.4	0.6	2	3	3	-38.5	78.9	9.8	6	II	47.4	66.6	90.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1768	NM_022005	Huesken	Fxyd6	53826	UCCCUCAAGCCUAAAUCCAga	UGGAUUUAGGCUUGAGGGA	-2.1	-2.4	1.6	2	0	3	-39.4	56.3	-5.7	5	II	47.4	53.3	81.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1769	NM_022005	Huesken	Fxyd6	53826	CCAGAAAGAGGGGACUAAGgg	CUUAGUCCCCUCUUUCUGG	-2.1	-3.3	0.3	-1	-1	4	-39.2	37.3	0.7	3	II	52.6	39.2	87.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1770	NM_022005	Huesken	Fxyd6	53826	CUAAGGGCUCCUCAGGGACgg	GUCCCUGAGGAGCCCUUAG	-2.2	-2.1	-4.4	-1	1	4	-44.1	43.7	8.5	2	II	63.2	47.7	85.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1771	NM_022005	Huesken	Fxyd6	53826	GGGAGACUCCAGAGAGCAAga	UUGCUCUCUGGAGUCUCCC	-0.9	-3.3	-3	0	-3	3	-43.1	29.3	-4	-1	III	57.9	16.6	58.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1772	NM_022005	Huesken	Fxyd6	53826	ACUCCAGAGAGCAAGAGUAga	UACUCUUGCUCUCUGGAGU	-1.3	-2.2	-3.7	3	0	2	-39.7	48.1	-8.7	4	II	47.4	42.9	57.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1773	NM_022005	Huesken	Fxyd6	53826	AGAGUAGAUGGAGGUCACAgg	UGUGACCUCCAUCUACUCU	-2.1	-2.1	0.8	0	0	2	-39.9	40.2	-3.8	5	II	47.4	43.4	96.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1774	NM_022005	Huesken	Fxyd6	53826	AUGGAGGUCACAGGAGGGCac	GCCCUCCUGUGACCUCCAU	-3.4	-1.1	0.9	1	4	4	-45.2	35	7.8	5	II	63.2	57.5	89	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1775	NM_022005	Huesken	Fxyd6	53826	AUGGGAGAGGAAGCACACAgu	UGUGUGCUUCCUCUCCCAU	-2.1	-1.1	2.5	2	0	3	-41.6	35.9	-1.5	4	II	52.6	48.3	104.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1776	NM_022005	Huesken	Fxyd6	53826	CACAGUUCACAGGAACAAUga	AUUGUUCCUGUGAACUGUG	-1.1	-2.1	-1.9	0	-3	2	-35.4	59.2	-3.7	2	II	42.1	40.6	76.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1777	NM_022005	Huesken	Fxyd6	53826	CAGUUCACAGGAACAAUGAaa	UCAUUGUUCCUGUGAACUG	-2.4	-2.1	-1.9	-1	-3	2	-35.6	63.3	-8.3	4	II	42.1	44.9	66.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1778	NM_022005	Huesken	Fxyd6	53826	UCACAGGAACAAUGAAACGac	CGUUUCAUUGUUCCUGUGA	-2.4	-2.4	-0.5	0	3	2	-34.7	56.3	5	6	II	42.1	64.1	54.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1779	NM_022005	Huesken	Fxyd6	53826	UGAAACGACCACGAACGACca	GUCGUUCGUGGUCGUUUCA	-2.2	-2.1	1	1	3	2	-38	52.6	10.1	7	Ib	52.6	60.2	39.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1780	XM_214061	Huesken	TCAP	8557	UUUGUUUUCAACAUGGUAAgg	UUACCAUGUUGAAAACAAA	-0.9	-0.9	0.4	2	1	2	-29.4	100.8	-4.1	9	II	26.3	70.8	91.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1781	XM_214061	Huesken	TCAP	8557	CAACAUGGUAAGGGUUUUGac	CAAAACCCUUACCAUGUUG	-2.1	-2.1	2	-1	0	3	-33.9	59.1	3	6	II	42.1	50.9	70.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1782	XM_214061	Huesken	TCAP	8557	CAUGGUAAGGGUUUUGACCuc	GGUCAAAACCCUUACCAUG	-3.3	-2.1	-0.9	0	2	3	-36.6	72.9	10.8	3	II	47.4	60.1	84.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1783	XM_214061	Huesken	TCAP	8557	AUGGUAAGGGUUUUGACCUca	AGGUCAAAACCCUUACCAU	-2.1	-1.1	-1.1	2	3	3	-36.6	70	1.4	4	II	42.1	63.9	85.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1784	XM_214061	Huesken	TCAP	8557	CCUCAAAUGCUUAUCAGGUcu	ACCUGAUAAGCAUUUGAGG	-2.2	-3.3	0.1	1	0	2	-35.8	57.2	-0.5	3	II	42.1	43.2	46.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1785	XM_214061	Huesken	TCAP	8557	CGCCUCUUCCUUGAAGUUUug	AAACUUCAAGGAAGAGGCG	-0.9	-2.4	-2.1	2	-4	4	-36.8	58.3	-3	2	III	47.4	37.9	66.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1786	XM_214061	Huesken	TCAP	8557	AUUGCAGGUUUUCCUUCUCag	GAGAAGGAAAACCUGCAAU	-2.4	-1.1	-1.4	1	4	2	-35.5	70.6	12.8	4	Ib	42.1	60.5	79.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1787	XM_214061	Huesken	TCAP	8557	ACACAAAGGGUAUUGCAAGga	CUUGCAAUACCCUUUGUGU	-2.1	-2.2	-1.6	3	3	3	-35.2	55.1	5	4	Ia	42.1	54.4	74.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1788	XM_214061	Huesken	TCAP	8557	CAAAGGGUAUUGCAAGGAUcu	AUCCUUGCAAUACCCUUUG	-1.1	-2.1	1.1	-1	-1	3	-35.5	50.3	-3.6	4	III	42.1	40.1	66.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1789	XM_214061	Huesken	TCAP	8557	CCUCAAAUAUAUGGGGAUAcc	UAUCCCCAUAUAUUUGAGG	-1.3	-3.3	0.8	-1	-4	4	-34.4	52.9	2.3	3	II	36.8	27.6	48.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1790	XM_214061	Huesken	TCAP	8557	UCAAAUAUAUGGGGAUACCaa	GGUAUCCCCAUAUAUUUGA	-3.3	-2.4	-0.4	-1	3	4	-34.5	70.8	10.1	7	Ia	36.8	67.1	76.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1791	XM_214061	Huesken	TCAP	8557	CAACUUGUACAGACGAACAag	UGUUCGUCUGUACAAGUUG	-2.1	-2.1	0.2	0	-2	2	-34.8	63.7	-6	4	II	42.1	49.5	84.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1792	XM_214061	Huesken	TCAP	8557	AACUUGUACAGACGAACAAga	UUGUUCGUCUGUACAAGUU	-0.9	-0.9	-0.7	2	0	2	-33.6	67.7	-1.4	7	II	36.8	48.3	55.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1793	XM_214061	Huesken	TCAP	8557	ACUUGUACAGACGAACAAGac	CUUGUUCGUCUGUACAAGU	-2.1	-2.2	-3.1	1	2	2	-34.8	62.2	-0.1	4	Ia	42.1	54.2	54.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1794	XM_214061	Huesken	TCAP	8557	UACAGACGAACAAGACCUCau	GAGGUCUUGUUCGUCUGUA	-2.4	-1.3	0.4	-1	3	2	-37.7	52.6	2.5	6	Ib	47.4	61.1	72.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1795	XM_214061	Huesken	TCAP	8557	GACCUCAUCUCCCAUACCGgg	CGGUAUGGGAGAUGAGGUC	-2.4	-2.4	1.6	2	2	3	-41.5	56.4	-2.3	2	II	57.9	50.6	84.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1796	XM_214061	Huesken	TCAP	8557	UCUCCCAUACCGGGUUCCGuu	CGGAACCCGGUAUGGGAGA	-2.4	-2.4	-3.7	0	4	5	-43.7	50.4	-2.4	3	II	63.2	56.1	49.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1797	XM_214061	Huesken	TCAP	8557	AAAAUGAGUUAGGCUGUCCca	GGACAGCCUAACUCAUUUU	-3.3	-0.9	-0.2	1	4	3	-35.7	72.2	9.7	8	Ia	42.1	70	62.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1798	XM_214061	Huesken	TCAP	8557	CAAACUCAUCAGGCAGGUUau	AACCUGCCUGAUGAGUUUG	-0.9	-2.1	1.9	-1	-1	3	-37.7	54.3	-3.7	3	II	47.4	40.1	52	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1799	XM_214061	Huesken	TCAP	8557	GCAGGUUAUAAUAGAUAGUgu	ACUAUCUAUUAUAACCUGC	-2.2	-3.4	2.8	0	-1	2	-32.2	64.2	-1.2	3	III	31.6	43.4	42.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1800	XM_214061	Huesken	TCAP	8557	AGUGUAGCACAUAAAAUUUaa	AAAUUUUAUGUGCUACACU	-0.9	-2.1	0.2	1	0	2	-29.8	75	-1.2	6	II	26.3	52.8	46.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1801	XM_214061	Huesken	TCAP	8557	AUACAUCUCUCCUCCUUUGgu	CAAAGGAGGAGAGAUGUAU	-2.1	-1.1	3.7	3	3	2	-36.3	75	-2	6	Ia	42.1	62.8	99.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1802	XM_214061	Huesken	TCAP	8557	UCCUCCUUUGGUCUCACACag	GUGUGAGACCAAAGGAGGA	-2.2	-2.4	-1	1	3	2	-40.5	81.2	15.2	4	II	52.6	61.8	98	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1803	XM_214061	Huesken	TCAP	8557	AGAAACAAUACAAUCCUCCga	GGAGGAUUGUAUUGUUUCU	-3.3	-2.1	0.5	1	4	2	-33.7	74.9	7.5	8	Ia	36.8	73.8	104.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1804	XM_214061	Huesken	TCAP	8557	AAACAAUACAAUCCUCCGAcc	UCGGAGGAUUGUAUUGUUU	-2.4	-0.9	1.5	4	2	3	-34	72.6	6.7	7	II	36.8	58.3	54.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1805	XM_214061	Huesken	TCAP	8557	UCCGACCCUUCUCCCAGGUca	ACCUGGGAGAAGGGUCGGA	-2.2	-2.4	-3.2	0	1	3	-45.5	47.4	1.8	2	II	63.2	45.8	77.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1806	XM_214061	Huesken	TCAP	8557	CCCUUCUCCCAGGUCAUCAgu	UGAUGACCUGGGAGAAGGG	-2.1	-3.3	-3.2	1	-3	3	-42.9	59	-1.1	2	III	57.9	38.5	33.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1807	XM_214061	Huesken	TCAP	8557	GUCCUAGCCGUGUAAGGCCca	GGCCUUACACGGCUAGGAC	-3.3	-2.2	-5	2	2	4	-43.3	43.6	7.4	1	II	63.2	49.2	60	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1808	XM_214061	Huesken	TCAP	8557	AUUGCUGCAUGGAGUCCUCuu	GAGGACUCCAUGCAGCAAU	-2.4	-1.1	0.5	0	4	2	-40.6	58	4.8	5	Ib	52.6	57.4	44.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1809	XM_214061	Huesken	TCAP	8557	GGUGAGCUCCAACUCCAAGuc	CUUGGAGUUGGAGCUCACC	-2.1	-3.3	-3.7	1	1	2	-41.4	47.5	9.7	0	II	57.9	36.3	55.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1810	XM_214061	Huesken	TCAP	8557	UCCAACUCCAAGUCUCAGUgu	ACUGAGACUUGGAGUUGGA	-2.2	-2.4	-0.1	2	0	2	-39.3	75.9	1	7	II	47.4	60.6	88.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1811	XM_214061	Huesken	TCAP	8557	AAGUCUCAGUGUCACAGGGac	CCCUGUGACACUGAGACUU	-3.3	-0.9	2.2	2	4	3	-40.3	65.5	5.4	4	Ib	52.6	56.6	65.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1812	XM_214061	Huesken	TCAP	8557	GUCUCAGUGUCACAGGGACcc	GUCCCUGUGACACUGAGAC	-2.2	-2.2	-1.5	1	1	3	-41.9	47.8	7.5	1	II	57.9	36.8	43	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1813	XM_214061	Huesken	TCAP	8557	AGUGUCACAGGGACCCAAAgu	UUUGGGUCCCUGUGACACU	-0.9	-2.1	-1.5	1	-1	3	-41.2	52	-6.3	4	II	52.6	42	50.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1814	XM_214061	Huesken	TCAP	8557	GUCACAGGGACCCAAAGUUca	AACUUUGGGUCCCUGUGAC	-0.9	-2.2	0.2	1	-1	3	-40	46.4	-6.3	4	III	52.6	37.9	26.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1815	XM_214061	Huesken	TCAP	8557	GCACCUCAAAUUCAGGAACac	GUUCCUGAAUUUGAGGUGC	-2.2	-3.4	-0.9	1	0	2	-36.8	57.8	12.8	2	II	47.4	41.2	23.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1816	XM_214061	Huesken	TCAP	8557	CUCAAAUUCAGGAACACACca	GUGUGUUCCUGAAUUUGAG	-2.2	-2.1	1.9	1	0	2	-34.4	69.6	5.5	5	II	42.1	58	32.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1817	XM_214061	Huesken	TCAP	8557	CCAGGAAUGAACGUUAUUCac	GAAUAACGUUCAUUCCUGG	-2.4	-3.3	-0.2	0	0	2	-34	61.6	7.7	4	II	42.1	52.4	66.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1818	XM_214061	Huesken	TCAP	8557	AUGUGAGCGCAUUUUCCAAau	UUGGAAAAUGCGCUCACAU	-0.9	-1.1	1	2	1	4	-35.8	68.6	3.9	5	II	42.1	45.6	38.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1819	XM_214061	Huesken	TCAP	8557	UGUGAGCGCAUUUUCCAAAuc	UUUGGAAAAUGCGCUCACA	-0.9	-2.1	1	1	0	4	-35.6	68.7	4.3	6	II	42.1	52.4	47.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1820	XM_214061	Huesken	TCAP	8557	CAUUUUCCAAAUCAAAGUUau	AACUUUGAUUUGGAAAAUG	-0.9	-2.1	1.4	-1	-1	2	-28.1	87.9	1.8	5	II	26.3	50.1	50.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1821	XM_214061	Huesken	TCAP	8557	UUCCAAAUCAAAGUUAUUUuc	AAAUAACUUUGAUUUGGAA	-0.9	-0.9	1.4	2	1	2	-27.3	90.4	3.4	8	II	21.1	69.3	70.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1822	XM_214061	Huesken	TCAP	8557	UCCAAAUCAAAGUUAUUUUcc	AAAAUAACUUUGAUUUGGA	-0.9	-2.4	1.4	1	0	2	-27.3	96	3.4	8	II	21.1	61.9	69.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1823	XM_214061	Huesken	TCAP	8557	AAACUCAUAUCGAGGUCGUca	ACGACCUCGAUAUGAGUUU	-2.2	-0.9	-0.9	1	3	2	-35.9	61.1	-8.9	6	II	42.1	53.3	70	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1824	XM_214061	Huesken	TCAP	8557	AGACCAUAGCAGAAACGUAgu	UACGUUUCUGCUAUGGUCU	-1.3	-2.1	0.5	4	0	2	-36.5	49.8	-4	4	II	42.1	43.5	62.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1825	XM_214061	Huesken	TCAP	8557	GAAGACAGAGGGGAAAAGAaa	UCUUUUCCCCUCUGUCUUC	-2.4	-2.4		0	-1	4	-38.2	42	-1.1	4	II	47.4	38.6	35.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1826	XM_214061	Huesken	TCAP	8557	GAAAACAAAGGCUCUCUGGuu	CCAGAGAGCCUUUGUUUUC	-3.3	-2.4	-2.8	0	2	3	-36.5	66.7	0.4	5	II	47.4	54.1	89.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1827	XM_214061	Huesken	TCAP	8557	AAGGCUCUCUGGUUUCAGCuu	GCUGAAACCAGAGAGCCUU	-3.4	-0.9	-1.4	2	4	3	-40.2	72.3	9.7	2	II	52.6	59.3	102	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1828	XM_214061	Huesken	TCAP	8557	UUUCAGCUUCUCCAGGGUGuc	CACCCUGGAGAAGCUGAAA	-2.1	-0.9	1.4	1	4	3	-40.1	58.3	2.8	6	Ib	52.6	64.9	66.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1829	XM_214061	Huesken	TCAP	8557	AGGGUGUCAUGUCAGCCAAuc	UUGGCUGACAUGACACCCU	-0.9	-2.1	0.8	2	-1	3	-41.4	52.8	1.7	4	II	52.6	39.6	65.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1830	XM_214061	Huesken	TCAP	8557	UCGCCAUCACUCAACCAGAau	UCUGGUUGAGUGAUGGCGA	-2.4	-2.4	0.5	2	-1	4	-41	49	-8.7	4	II	52.6	52.5	75.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1831	XM_214061	Huesken	TCAP	8557	UGGAAACCGCGGACGGCUCac	GAGCCGUCCGCGGUUUCCA	-2.4	-2.1	-6.3	1	2	6	-44.8	41.5	7.5	3	Ib	68.4	53.6	70.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1832	XM_214061	Huesken	TCAP	8557	GAAACCGCGGACGGCUCACga	GUGAGCCGUCCGCGGUUUC	-2.2	-2.4	-6.5	0	2	6	-43.7	37.7	7.4	2	II	68.4	40.1	72.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1833	XM_214061	Huesken	TCAP	8557	CGCGGACGGCUCACGAAGAcu	UCUUCGUGAGCCGUCCGCG	-2.4	-2.4	-2	0	-4	5	-44.2	19.5	-10.7	-2	III	68.4	24	35.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1834	XM_214061	Huesken	TCAP	8557	AUCCGCCGCACGAGCUGCAcc	UGCAGCUCGUGCGGCGGAU	-2.1	-1.1	-1.9	2	1	7	-46	32.3	-8.7	2	II	68.4	41.1	33.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1835	XM_214061	Huesken	TCAP	8557	CGAGCUGCACCAGCAUCUCgg	GAGAUGCUGGUGCAGCUCG	-2.4	-2.4	-3.1	-1	0	2	-43.2	37.2	3.1	0	II	63.2	42.5	32.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1836	XM_214061	Huesken	TCAP	8557	CCAGCAUCUCGGCCAUCUUcg	AAGAUGGCCGAGAUGCUGG	-0.9	-3.3	-1.3	0	-2	5	-42	43.9	-3.6	0	III	57.9	34	33.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1837	XM_214061	Huesken	TCAP	8557	GGCCAUCUUCGCCAUCUCCug	GGAGAUGGCGAAGAUGGCC	-3.3	-3.3	-2.4	1	0	4	-43.5	48.4	10.2	0	II	63.2	40.7	72.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1838	XM_214061	Huesken	TCAP	8557	CCAUCUUCGCCAUCUCCUGcg	CAGGAGAUGGCGAAGAUGG	-2.1	-3.3	1.1	1	0	4	-41	57	-4.4	2	II	57.9	47.1	81.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1839	XM_214061	Huesken	TCAP	8557	CGCCCUGCAGGUGGUUGAGag	CUCAACCACCUGCAGGGCG	-2.1	-2.4	-1.2	-1	-1	5	-44.8	36.7	1	-1	II	68.4	27.2	43.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1840	XM_214061	Huesken	TCAP	8557	GAGAGCGCUCUCCUCAGAGgc	CUCUGAGGAGAGCGCUCUC	-2.1	-2.4	-2.5	0	2	4	-43.7	46	2.4	1	II	63.2	43.3	48.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1841	XM_214061	Huesken	TCAP	8557	GCUCUCCUCAGAGGCCUCUac	AGAGGCCUCUGAGGAGAGC	-2.1	-3.4	-2.4	1	-1	4	-45.5	40.8	-4	2	III	63.2	32.5	69.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1842	XM_214061	Huesken	TCAP	8557	ACCUGCGUCUUGAUCUGGAuc	UCCAGAUCAAGACGCAGGU	-2.4	-2.2	1.3	2	1	3	-41	52.6	-4	4	II	52.6	41.1	54.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1843	XM_214061	Huesken	TCAP	8557	CAUAUUGUCCAUCUCCCACag	GUGGGAGAUGGACAAUAUG	-2.2	-2.1	1.4	0	2	3	-37.5	80.3	15.5	5	II	47.4	57	79.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1844	XM_214061	Huesken	TCAP	8557	UCUCCCACAGCUUGCCCCGgu	CGGGGCAAGCUGUGGGAGA	-2.4	-2.4	-0.1	0	4	6	-46.2	52.8	0.7	4	II	68.4	60.8	69.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1845	XM_214061	Huesken	TCAP	8557	CUCCCACAGCUUGCCCCGGuu	CCGGGGCAAGCUGUGGGAG	-3.3	-2.1	-0.4	1	1	7	-47.1	30.5	1	0	II	73.7	39.9	44.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1846	XM_214061	Huesken	TCAP	8557	GCGCACGGGCGCGCGUCCGcg	CGGACGCGCGCCCGUGCGC	-2.4	-3.4	-7	2	1	10	-50.2	12.2	0.4	0	II	89.5	31.4	19.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1847	XM_214061	Huesken	TCAP	8557	CGCGCUCGCGCGUCAGCGAcg	UCGCUGACGCGCGAGCGCG	-2.4	-2.4	-5.8	0	-3	6	-47.1	25.2	-8.1	-2	III	78.9	23.7	14	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1848	XM_214061	Huesken	TCAP	8557	UCUCGUGGAUGACGGGCAGca	CUGCCCGUCAUCCACGAGA	-2.1	-2.4	-0.5	0	3	5	-43.7	40.8	2.4	4	II	63.2	47.3	70.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1849	XM_214061	Huesken	TCAP	8557	AUGACGGGCAGCAGAAACUgc	AGUUUCUGCUGCCCGUCAU	-2.1	-1.1	0	1	1	5	-40.5	52.2	-8.9	6	II	52.6	57.5	66.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1850	XM_214061	Huesken	TCAP	8557	AGCAGAAACUGCUCGAAGUcc	ACUUCGAGCAGUUUCUGCU	-2.2	-2.1	-3.9	2	0	2	-38.2	59.6	-6.3	3	II	47.4	47.1	51.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1851	XM_214061	Huesken	TCAP	8557	AGCACCUCCACCUCCUCGGgc	CCGAGGAGGUGGAGGUGCU	-3.3	-2.1	-0.7	3	3	3	-46.5	60.5	-4.7	4	II	68.4	57.5	58.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1852	XM_214061	Huesken	TCAP	8557	GCCGCAUCUCCGCUGCUUUgc	AAAGCAGCGGAGAUGCGGC	-0.9	-3.4	-2.6	1	-2	5	-43.2	44.4	-1.6	0	III	63.2	28.6	35.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1853	XM_214061	Huesken	TCAP	8557	CGCAUCUCCGCUGCUUUGCuu	GCAAAGCAGCGGAGAUGCG	-3.4	-2.4	-2.6	1	-1	4	-42	59.8	6.1	2	II	63.2	46.4	74.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1854	XM_214061	Huesken	TCAP	8557	UCUCCGCUGCUUUGCUUGUcu	ACAAGCAAAGCAGCGGAGA	-2.2	-2.4	-1.4	1	1	4	-40.3	60.5	-1	4	II	52.6	49	62.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1855	XM_214061	Huesken	TCAP	8557	UAUUAACAUGUAAGAUUUUgu	AAAAUCUUACAUGUUAAUA	-0.9	-1.3	2.6	-1	1	1	-25.7	80.2	-3.9	8	II	15.8	60.6	76.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1856	XM_214061	Huesken	TCAP	8557	UAAGGGUUUUGACCUCAAAug	UUUGAGGUCAAAACCCUUA	-0.9	-1.3	-2.5	1	0	3	-34.2	68.6	-1.4	5	II	36.8	56.6	82.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1857	XM_214061	Huesken	TCAP	8557	ACAUUUAUAGCUUGACUACau	GUAGUCAAGCUAUAAAUGU	-2.2	-2.2	-1.6	0	2	2	-31.7	83.3	8.1	6	Ia	31.6	58.7	88.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1858	XM_214061	Huesken	TCAP	8557	UAUAGCUUGACUACAUAUGaa	CAUAUGUAGUCAAGCUAUA	-2.1	-1.3	-2.2	2	3	2	-32.2	91.5	-1.9	7	Ia	31.6	76.1	101.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1859	XM_214061	Huesken	TCAP	8557	UUGACUACAUAUGAACCUUga	AAGGUUCAUAUGUAGUCAA	-0.9	-0.9	1.4	0	1	2	-32.7	81.4	1.4	6	II	31.6	66.3	85.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1860	XM_214061	Huesken	TCAP	8557	GACGCCUCUUCCUUGAAGUuu	ACUUCAAGGAAGAGGCGUC	-2.2	-2.4	-2.1	1	0	4	-39.6	64.1	-6.3	1	III	52.6	41.2	82.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1861	XM_214061	Huesken	TCAP	8557	GUUUUGGGUCCAUUGCAGGuu	CCUGCAAUGGACCCAAAAC	-3.3	-2.2	-1	0	3	3	-38.6	62.3	0	4	II	52.6	51.5	86.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1862	XM_214061	Huesken	TCAP	8557	GGUAUUGCAAGGAUCUUUCau	GAAAGAUCCUUGCAAUACC	-2.4	-3.3	2.1	0	1	2	-34.9	78	7.1	6	II	42.1	52.9	82.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1863	XM_214061	Huesken	TCAP	8557	AUAUGGGGAUACCAACUUGua	CAAGUUGGUAUCCCCAUAU	-2.1	-1.1	-1.1	1	3	4	-36.1	62.9	2.4	6	Ib	42.1	61.3	62.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1864	XM_214061	Huesken	TCAP	8557	GAGUUAGGCUGUCCCAACAaa	UGUUGGGACAGCCUAACUC	-2.1	-2.4	-6.4	1	-2	3	-40.4	57.1	1.7	3	II	52.6	38	85.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1865	XM_214061	Huesken	TCAP	8557	AUCAGGCAGGUUAUAAUAGau	CUAUUAUAACCUGCCUGAU	-2.1	-1.1	3	2	3	3	-34.1	68.7	2.8	5	II	36.8	58.8	88	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1866	XM_214061	Huesken	TCAP	8557	GGCAGGUUAUAAUAGAUAGug	CUAUCUAUUAUAACCUGCC	-2.1	-3.3	2.8	1	-1	3	-33.3	70.3	2.3	3	II	36.8	42.6	90.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1867	XM_214061	Huesken	TCAP	8557	CAUAGAAACAAUACAAUCCuc	GGAUUGUAUUGUUUCUAUG	-3.3	-2.1	0.7	0	1	2	-30.4	82.7	8.1	6	II	31.6	67.3	93.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1868	XM_214061	Huesken	TCAP	8557	UCCCAGGUCAUCAGUCCUAgc	UAGGACUGAUGACCUGGGA	-1.3	-2.4	-2.2	1	-1	3	-42.2	43.5	-6	5	II	52.6	49	70.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1869	XM_214061	Huesken	TCAP	8557	CCCAGGUCAUCAGUCCUAGcc	CUAGGACUGAUGACCUGGG	-2.1	-3.3	-5.3	0	-1	3	-41.9	45.2	-2	1	II	57.9	38.3	63.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1870	XM_214061	Huesken	TCAP	8557	CAUCAGUCCUAGCCGUGUAag	UACACGGCUAGGACUGAUG	-1.3	-2.1	-0.8	2	-2	4	-40.1	54	1.7	4	II	52.6	37.1	61.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1871	XM_214061	Huesken	TCAP	8557	CACAGUCCCAGAUGAAGUCug	GACUUCAUCUGGGACUGUG	-2.4	-2.1	0.7	-1	0	3	-39.5	56	5.4	3	II	52.6	47.2	82.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1872	XM_214061	Huesken	TCAP	8557	UUCUUCUGCCAUUGCUGCAug	UGCAGCAAUGGCAGAAGAA	-2.1	-0.9	-1.3	1	1	3	-38.9	79.4	1.6	6	II	47.4	59.3	72.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1873	XM_214061	Huesken	TCAP	8557	CUUCUGCCAUUGCUGCAUGga	CAUGCAGCAAUGGCAGAAG	-2.1	-2.1	-1.6	0	1	3	-38.8	57.2	2.7	2	II	52.6	47	79.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1874	XM_214061	Huesken	TCAP	8557	UGCAUGGAGUCCUCUUCUAga	UAGAAGAGGACUCCAUGCA	-1.3	-2.1	0.5	1	-1	2	-39.8	61	-8.5	6	II	47.4	48.1	68.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1875	XM_214061	Huesken	TCAP	8557	GAACUUUCAGAGGCGGUGAgc	UCACCGCCUCUGAAAGUUC	-2.4	-2.4	0.9	2	-1	5	-39.6	56.2	-1.5	5	II	52.6	41.6	45.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1876	XM_214061	Huesken	TCAP	8557	GGCGGUGAGCUCCAACUCCaa	GGAGUUGGAGCUCACCGCC	-3.3	-3.3	-3.7	1	0	5	-45.4	32	7.5	0	II	68.4	42.7	50	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1877	XM_214061	Huesken	TCAP	8557	CUCCAACUCCAAGUCUCAGug	CUGAGACUUGGAGUUGGAG	-2.1	-2.1	-0.1	1	1	2	-39.2	50.9	5.3	2	II	52.6	40.8	81.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1878	XM_214061	Huesken	TCAP	8557	CCAACUCCAAGUCUCAGUGuc	CACUGAGACUUGGAGUUGG	-2.1	-3.3	-0.1	-1	1	2	-39	66.6	5.8	3	II	52.6	47.7	77.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1879	XM_214061	Huesken	TCAP	8557	GGGACCCAAAGUUCAGGGAgc	UCCCUGAACUUUGGGUCCC	-2.4	-3.3	-3.7	0	-2	3	-42.5	50.9	-0.7	0	III	57.9	26.1	53.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1880	XM_214061	Huesken	TCAP	8557	GUUCAGGGAGCUAUCAGGUca	ACCUGAUAGCUCCCUGAAC	-2.2	-2.2	1.2	1	2	3	-40.8	49.4	-0.9	4	III	52.6	43.7	68.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1881	XM_214061	Huesken	TCAP	8557	GGGAGCUAUCAGGUCACCCaa	GGGUGACCUGAUAGCUCCC	-3.3	-3.3	-4.3	1	2	3	-44.4	48.9	12	0	II	63.2	48.5	87.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1882	XM_214061	Huesken	TCAP	8557	GCUAUCAGGUCACCCAAGAaa	UCUUGGGUGACCUGAUAGC	-2.4	-3.4	-4.3	1	-2	3	-40.8	47.9	-3.7	3	II	52.6	32.8	45.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1883	XM_214061	Huesken	TCAP	8557	AGGUCACCCAAGAAAGAGCaa	GCUCUUUCUUGGGUGACCU	-3.4	-2.1	-4.3	2	2	3	-40.2	57.5	7.1	3	II	52.6	54.4	95.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1884	XM_214061	Huesken	TCAP	8557	GAACACACCAAAGCCAGGAau	UCCUGGCUUUGGUGUGUUC	-2.4	-2.4	-0.1	2	0	3	-40	46.7	1.3	5	II	52.6	43.3	66.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1885	XM_214061	Huesken	TCAP	8557	ACACCAAAGCCAGGAAUGAac	UCAUUCCUGGCUUUGGUGU	-2.4	-2.2	-0.1	2	0	3	-38.9	36.5	-6.1	5	II	47.4	38.5	52.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1886	XM_214061	Huesken	TCAP	8557	GUAUGUGAGCGCAUUUUCCaa	GGAAAAUGCGCUCACAUAC	-3.3	-2.2	0.6	1	3	4	-36.3	68.3	4.8	5	II	47.4	57.5	86.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1887	XM_214061	Huesken	TCAP	8557	AGCGCAUUUUCCAAAUCAAag	UUGAUUUGGAAAAUGCGCU	-0.9	-2.1	1.5	2	-1	4	-33.3	59.1	-11	3	II	36.8	39.4	68.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1888	XM_214061	Huesken	TCAP	8557	CCAUAGCAGAAACGUAGUGuc	CACUACGUUUCUGCUAUGG	-2.1	-3.3	0.5	-1	0	2	-36.2	47.4	5.4	4	II	47.4	40.7	58	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1889	XM_214061	Huesken	TCAP	8557	AGCAGAAACGUAGUGUCAGug	CUGACACUACGUUUCUGCU	-2.1	-2.1	0.8	2	3	2	-37.2	52.8	2.3	4	Ib	47.4	51.2	99.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1890	XM_214061	Huesken	TCAP	8557	AACGUAGUGUCAGUGAGAAca	UUCUCACUGACACUACGUU	-0.9	-0.9	0.9	2	1	2	-36.3	54.1	-3.8	5	II	42.1	40.9	72.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1891	XM_214061	Huesken	TCAP	8557	CGUAGUGUCAGUGAGAACAga	UGUUCUCACUGACACUACG	-2.1	-2.4	-0.2	0	-3	2	-37.5	61.4	-2.7	4	III	47.4	44.9	68.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1892	XM_214061	Huesken	TCAP	8557	CAGAGUGAAGACAGAGGGGaa	CCCCUCUGUCUUCACUCUG	-3.3	-2.1	1.9	-2	1	4	-41.6	40.8	-4.6	2	II	57.9	47.1	87.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1893	XM_214061	Huesken	TCAP	8557	AAGGAAAACAAAGGCUCUCug	GAGAGCCUUUGUUUUCCUU	-2.4	-0.9	-3.4	1	3	3	-35.3	55.9	12.4	5	Ia	42.1	62.2	106.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1894	XM_214061	Huesken	TCAP	8557	ACAAAGGCUCUCUGGUUUCag	GAAACCAGAGAGCCUUUGU	-2.4	-2.2	-1.3	0	2	3	-37.8	57	7.8	6	Ib	47.4	56.6	86	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1895	XM_214061	Huesken	TCAP	8557	UCAGCCAAUCGCCAUCACUca	AGUGAUGGCGAUUGGCUGA	-2.1	-2.4	-1.9	0	1	4	-40.9	57.6	-3.9	4	II	52.6	59.5	85.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1896	XM_214061	Huesken	TCAP	8557	UUCUCUAUCCGCCGCACGAgc	UCGUGCGGCGGAUAGAGAA	-2.4	-0.9	-0.8	0	1	7	-42	63.8	-3.7	5	II	57.9	56.8	87.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1897	XM_214061	Huesken	TCAP	8557	UCUAUCCGCCGCACGAGCUgc	AGCUCGUGCGGCGGAUAGA	-2.1	-2.4	-1.9	1	2	7	-44.2	54.7	-8.9	4	II	63.2	54.7	79.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1898	XM_214061	Huesken	TCAP	8557	CCGCCGCACGAGCUGCACCag	GGUGCAGCUCGUGCGGCGG	-3.3	-3.3	-2.3	1	0	7	-48	32.4	12.4	-2	II	78.9	40.2	88.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1899	XM_214061	Huesken	TCAP	8557	UCUCGGCCAUCUUCGCCAUcu	AUGGCGAAGAUGGCCGAGA	-1.1	-2.4	-4.4	1	1	5	-42.6	51.4	-3.3	2	II	57.9	43	78.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1900	XM_214061	Huesken	TCAP	8557	GCGCCUUCUCGUCGGCCUUcu	AAGGCCGACGAGAAGGCGC	-0.9	-3.4	-5	1	-2	5	-44.8	39.2	1.5	-1	III	68.4	27.5	63.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1901	XM_214061	Huesken	TCAP	8557	CGUCGGCCUUCUCCGCCCUgg	AGGGCGGAGAAGGCCGACG	-2.1	-2.4	-4.1	0	-1	6	-47.1	31.1	-2.9	-1	III	73.7	31.3	23.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1902	XM_214061	Huesken	TCAP	8557	CGGCCUUCUCCGCCCUGGGgc	CCCAGGGCGGAGAAGGCCG	-3.3	-2.4	-4.1	1	0	6	-48.8	37	-2	-1	II	78.9	40	65.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1903	XM_214061	Huesken	TCAP	8557	GCGGGGCUCGGCGCCGCCCgc	GGGCGGCGCCGAGCCCCGC	-3.3	-3.4	-7.4	1	1	11	-54.2	9.4	5.1	-2	II	94.7	34.9	74.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1904	XM_214061	Huesken	TCAP	8557	GCCGCCCGCGCCCUGCAGGug	CCUGCAGGGCGCGGGCGGC	-3.3	-3.4	-6.5	1	1	13	-52.4	22.1	0	-2	II	89.5	28.8	60.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1905	XM_214061	Huesken	TCAP	8557	GGCCUCUACCUGCGUCUUGau	CAAGACGCAGGUAGAGGCC	-2.1	-3.3	-1	2	-1	4	-43.1	44	2.4	1	II	63.2	34.9	73.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1906	XM_214061	Huesken	TCAP	8557	CUCUACCUGCGUCUUGAUCug	GAUCAAGACGCAGGUAGAG	-2.4	-2.1	0.6	0	0	3	-39	68.7	15.9	3	II	52.6	44.5	72	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1907	XM_214061	Huesken	TCAP	8557	UGGAUCAGCAUAUUGUCCAuc	UGGACAAUAUGCUGAUCCA	-2.1	-2.1	-0.9	1	1	2	-37.5	76.8	-1.4	7	II	42.1	65	91.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1908	XM_214061	Huesken	TCAP	8557	UAUUGUCCAUCUCCCACAGcu	CUGUGGGAGAUGGACAAUA	-2.1	-1.3	0.6	0	4	3	-38.5	76.3	3.1	6	Ia	47.4	62.5	82.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1909	XM_214061	Huesken	TCAP	8557	CCAUCUCCCACAGCUUGCCcc	GGCAAGCUGUGGGAGAUGG	-3.3	-3.3	0.7	0	1	3	-43.7	50.4	3.1	2	II	63.2	47.4	81	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1910	XM_214061	Huesken	TCAP	8557	CCCGGUUGCGCACGGGCGCgc	GCGCCCGUGCGCAACCGGG	-3.4	-3.3	-6.1	0	0	7	-49.2	19.4	5.5	-1	II	84.2	37.3	55.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1911	XM_214061	Huesken	TCAP	8557	UCAGCGACGCGAUGUCCUCgc	GAGGACAUCGCGUCGCUGA	-2.4	-2.4	-2.2	0	3	4	-43.2	45.1	7.4	3	II	63.2	58.6	77.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1912	XM_214061	Huesken	TCAP	8557	CGUGGAUGACGGGCAGCAGaa	CUGCUGCCCGUCAUCCACG	-2.1	-2.4	0.6	0	0	5	-44.4	29.8	-2	1	II	68.4	32.2	39.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1913	XM_214061	Huesken	TCAP	8557	GUGGAUGACGGGCAGCAGAaa	UCUGCUGCCCGUCAUCCAC	-2.4	-2.2	1.9	1	-2	5	-44.4	36	1.4	2	III	63.2	37	35.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1914	XM_214061	Huesken	TCAP	8557	CAGCAGAAACUGCUCGAAGuc	CUUCGAGCAGUUUCUGCUG	-2.1	-2.1	-4.6	0	0	2	-38.1	48.8	5.4	2	II	52.6	46.5	59.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1915	XM_214061	Huesken	TCAP	8557	UCCGGCUCCAGCACCUCCAcc	UGGAGGUGCUGGAGCCGGA	-2.1	-2.4	-3.1	2	0	5	-47.7	47.3	-8.7	3	II	68.4	54.8	69.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1916	XM_214061	Huesken	TCAP	8557	UCCGCCAGCUCCACUGCGUcc	ACGCAGUGGAGCUGGCGGA	-2.2	-2.4	-2.5	1	1	5	-46.7	42.2	-11.3	2	II	68.4	48.9	76.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1917	XM_214061	Huesken	TCAP	8557	CCGCCAGCUCCACUGCGUCca	GACGCAGUGGAGCUGGCGG	-2.4	-3.3	-2.5	0	0	5	-46.7	29.7	7.8	-1	II	73.7	35.8	65.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1918	XM_214061	Huesken	TCAP	8557	GGCCGCAUCUCCGCUGCUUug	AAGCAGCGGAGAUGCGGCC	-0.9	-3.3	-2.6	2	-2	6	-45.6	31.1	-8.7	-1	III	68.4	25.4	42.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1919	XM_214061	Huesken	TCAP	8557	GCUUUGCUUGUCUGCACGAca	UCGUGCAGACAAGCAAAGC	-2.4	-3.4	-2.1	-1	-1	2	-39.2	48.5	-6.1	2	III	52.6	31.4	67.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1920	XM_214061	Huesken	TCAP	8557	GCUUGUCUGCACGACACUCag	GAGUGUCGUGCAGACAAGC	-2.4	-3.4	-3.8	1	1	2	-40.8	48.3	7.4	0	II	57.9	41.1	64.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1921	XM_214061	Huesken	TCAP	8557	GACCGAGCUCGGGCCGAGUcu	ACUCGGCCCGAGCUCGGUC	-2.2	-2.4	-3.1	1	-1	7	-47.5	23.6	-4	0	III	73.7	28.8	61.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1922	XM_214061	Huesken	TCAP	8557	GUCUUCAAAACUUCCGCCGcu	CGGCGGAAGUUUUGAAGAC	-2.4	-2.2	-2.5	0	2	6	-37.2	72	0.7	4	II	52.6	56.3	94.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1923	XM_214061	Huesken	TCAP	8557	AAAACUUCCGCCGCUGUUCca	GAACAGCGGCGGAAGUUUU	-2.4	-0.9	-2.5	3	3	7	-38.2	70.4	2.7	6	Ia	52.6	67.2	84.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1924	NM_017346	Huesken	Cacnb1	782	CUGCCUCUCAGCGGAUGUAga	UACAUCCGCUGAGAGGCAG	-1.3	-2.1	-5.2	0	-3	4	-42.3	42.7	-8.1	0	III	57.9	30.8	39.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1925	NM_017346	Huesken	Cacnb1	782	CAGCCCUCCAGCUCAUUCUua	AGAAUGAGCUGGAGGGCUG	-2.1	-2.1	-2.9	2	-2	4	-42.7	64.8	-8.3	2	III	57.9	48.7	75.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1926	NM_017346	Huesken	Cacnb1	782	AGCCCUCCAGCUCAUUCUUau	AAGAAUGAGCUGGAGGGCU	-0.9	-2.1	-2.2	2	0	4	-41.5	56.8	-0.9	1	II	52.6	40.1	54.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1927	NM_017346	Huesken	Cacnb1	782	UUGCGCCCCAGAACCGGCCca	GGCCGGUUCUGGGGCGCAA	-3.3	-0.9	-4.4	2	4	7	-47.5	45.1	4.8	3	II	73.7	67.7	79.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1928	NM_017346	Huesken	Cacnb1	782	GGCCCACCACCCUCCGCACag	GUGCGGAGGGUGGUGGGCC	-2.2	-3.3	-0.5	1	0	5	-49.7	29.1	2.7	-1	II	78.9	25.2	52.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1929	NM_017346	Huesken	Cacnb1	782	CCUCCUCUUCCCAGGAGCCcu	GGCUCCUGGGAAGAGGAGG	-3.3	-3.3	-6.7	-1	1	3	-46.3	39.1	-1.9	-1	II	68.4	41.2	46.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1930	NM_017346	Huesken	Cacnb1	782	GAGUCUCCUGGCUCCGUGUag	ACACGGAGCCAGGAGACUC	-2.2	-2.4	-1.8	0	-1	3	-44.5	56.8	-6.3	1	III	63.2	38.4	48.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1931	NM_017346	Huesken	Cacnb1	782	AGUCUCCUGGCUCCGUGUAga	UACACGGAGCCAGGAGACU	-1.3	-2.1	-1.7	3	0	3	-43.4	52.7	-0.7	2	II	57.9	35.6	54.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1932	NM_017346	Huesken	Cacnb1	782	CCGUGUAGACAGAGUUUCGgc	CGAAACUCUGUCUACACGG	-2.4	-3.3	-0.3	-2	-1	3	-37.7	52.6	-2.3	3	II	52.6	44.5	44	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1933	NM_017346	Huesken	Cacnb1	782	UCAAAGGUGUCUUGGCGGGau	CCCGCCAAGACACCUUUGA	-3.3	-2.4	0.1	0	4	6	-41.4	51.7	3.4	6	Ib	57.9	58.4	57.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1934	NM_017346	Huesken	Cacnb1	782	AGGUGUCUUGGCGGGAUAGcg	CUAUCCCGCCAAGACACCU	-2.1	-2.1	2.2	0	1	6	-42	49.4	-2.4	3	II	57.9	43.2	70.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1935	NM_017346	Huesken	Cacnb1	782	GGCCAGGUGGGUGGUUGCUgg	AGCAACCACCCACCUGGCC	-2.1	-3.3	-1.1	1	-1	4	-46.7	26.9	1.8	0	III	68.4	24.5	31.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1936	NM_017346	Huesken	Cacnb1	782	GGGGCUGGCCCAGCUCCCCgg	GGGGAGCUGGGCCAGCCCC	-3.3	-3.3	-4.6	1	1	5	-52.8	21.5	5.1	-1	II	84.2	39.2	19.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1937	NM_017346	Huesken	Cacnb1	782	GGUAGCCAUGGUGCGGUUCag	GAACCGCACCAUGGCUACC	-2.4	-3.3	-0.9	0	0	4	-43.1	47.5	10.4	2	II	63.2	36.4	49	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1938	NM_017346	Huesken	Cacnb1	782	UCAGCAGCGGAUUGGGUGGcg	CCACCCAAUCCGCUGCUGA	-3.3	-2.4	2.1	0	3	4	-44	51	5.3	5	II	63.2	56.9	81.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1939	NM_017346	Huesken	Cacnb1	782	CUUCCAGUAGGCUUCCAAGua	CUUGGAAGCCUACUGGAAG	-2.1	-2.1	-1	0	1	3	-39	63.2	-2	2	II	52.6	47	60.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1940	NM_017346	Huesken	Cacnb1	782	AGGCAUCUUCCAAUUGGUUcu	AACCAAUUGGAAGAUGCCU	-0.9	-2.1	-1.1	2	1	3	-36.6	57.8	-3.9	2	II	42.1	44.9	50	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1941	NM_017346	Huesken	Cacnb1	782	CAAUUGGUUCUCGUCCAGGau	CCUGGACGAGAACCAAUUG	-3.3	-2.1	-1	-1	2	2	-38.2	57.8	-2	4	II	52.6	57.1	80.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1942	NM_017346	Huesken	Cacnb1	782	UGAUGAGCCUCUGCAGUACcu	GUACUGCAGAGGCUCAUCA	-2.2	-2.1	-0.4	1	2	3	-40.9	58.1	8.1	7	Ib	52.6	55.5	73.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1943	NM_017346	Huesken	Cacnb1	782	AGCGAGGUUUUAGAGAGCUgg	AGCUCUCUAAAACCUCGCU	-2.1	-2.1	0.4	1	1	3	-38.5	43.5	-1.3	4	II	47.4	49.4	56.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1944	NM_017346	Huesken	Cacnb1	782	GAUCCGCUCGAUUUCACUCug	GAGUGAAAUCGAGCGGAUC	-2.4	-2.4	0.2	2	3	4	-38.4	70.4	15.1	2	II	52.6	50.8	84.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1945	NM_017346	Huesken	Cacnb1	782	AUUUCACUCUGUACCUCAGcc	CUGAGGUACAGAGUGAAAU	-2.1	-1.1	-0.8	2	4	2	-36	78.7	0.4	5	Ia	42.1	55.1	59.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1946	NM_017346	Huesken	Cacnb1	782	CACUCUGUACCUCAGCCAGgc	CUGGCUGAGGUACAGAGUG	-2.1	-2.1	-1.3	-1	0	3	-41.6	60.1	1.1	1	II	57.9	44	34.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1947	NM_017346	Huesken	Cacnb1	782	GCGAGUGUUGGAGCGCUCGau	CGAGCGCUCCAACACUCGC	-2.4	-3.4	-1.9	0	0	4	-43.7	37.9	0	1	II	68.4	43.5	15.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1948	NM_017346	Huesken	Cacnb1	782	AUAAUGUGUUUGCUGGGGUug	ACCCCAGCAAACACAUUAU	-2.2	-1.1	3.9	2	4	4	-36.6	70.3	0.8	7	II	42.1	61.1	63.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1949	NM_017346	Huesken	Cacnb1	782	UUGAGCGAUGGUCCCACCAgg	UGGUGGGACCAUCGCUCAA	-2.1	-0.9	-4	0	1	3	-43.1	58.4	1.7	4	II	57.9	61.8	71.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1950	NM_017346	Huesken	Cacnb1	782	GGGGCACGUGCUCUGUCGAcu	UCGACAGAGCACGUGCCCC	-2.4	-3.3	-0.6	0	-1	5	-45.8	33.9	1.7	-1	III	68.4	22	39.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1951	NM_017346	Huesken	Cacnb1	782	UUCUGCUUCUGUUUGGCACug	GUGCCAAACAGAAGCAGAA	-2.2	-0.9	-0.6	2	4	3	-37.6	89.7	12.8	6	Ib	47.4	66.9	82.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1952	NM_017346	Huesken	Cacnb1	782	UGCUGGAGCUGAGGCGGUUuu	AACCGCCUCAGCUCCAGCA	-0.9	-2.1	-1.4	0	0	5	-45.1	35.7	-4	3	II	63.2	42.7	46.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1953	NM_017346	Huesken	Cacnb1	782	GAGCUGAGGCGGUUUUGGCgc	GCCAAAACCGCCUCAGCUC	-3.4	-2.4	-0.3	2	3	5	-42.7	51.1	9.7	2	II	63.2	45.7	54.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1954	NM_017346	Huesken	Cacnb1	782	AGCAGGCGAAGGCUGUCCAgu	UGGACAGCCUUCGCCUGCU	-2.1	-2.1	-0.5	1	1	4	-45.1	45.6	1	3	II	63.2	40.1	39.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1955	NM_017346	Huesken	Cacnb1	782	GACAGGGCUGGGGAUGAAAcc	UUUCAUCCCCAGCCCUGUC	-0.9	-2.4	1.4	0	-3	4	-42.7	35.1	-3.8	2	III	57.9	27.2	16.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1956	NM_017346	Huesken	Cacnb1	782	UCCUUCACCAGCCUCCCGAuc	UCGGGAGGCUGGUGAAGGA	-2.4	-2.4	-0.9	1	1	4	-45.5	61.6	-1.4	6	II	63.2	54.6	73.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1957	NM_017346	Huesken	Cacnb1	782	CAGCCUCCCGAUCCACCAGuc	CUGGUGGAUCGGGAGGCUG	-2.1	-2.1	-1	1	0	4	-45.4	46.7	5.7	-1	II	68.4	40.7	40.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1958	NM_017346	Huesken	Cacnb1	782	UUCUCCUUGAUGUGCAGGAag	UCCUGCACAUCAAGGAGAA	-2.4	-0.9	-3.9	1	1	2	-39.4	73.7	-3.8	6	II	47.4	59.7	94	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1959	NM_017346	Huesken	Cacnb1	782	AUGUGCAGGAAGUCCUUGGgc	CCAAGGACUUCCUGCACAU	-3.3	-1.1	-2.4	2	3	2	-40.1	70.1	2.3	6	Ib	52.6	64.7	63.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1960	NM_017346	Huesken	Cacnb1	782	ACUGGCACCUCAUCUCCUGga	CAGGAGAUGAGGUGCCAGU	-2.1	-2.2	0	2	3	3	-42.8	51	-2.4	3	II	57.9	52.6	83.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1961	NM_017346	Huesken	Cacnb1	782	ACGGAUUGUAGCCAACAUUug	AAUGUUGGCUACAAUCCGU	-0.9	-2.2	-1	2	-1	3	-36	59.5	-1.2	4	II	42.1	52.6	90.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1962	NM_017346	Huesken	Cacnb1	782	CUUUCUCGAGCUGGGCUAAgg	UUAGCCCAGCUCGAGAAAG	-0.9	-2.1	0.4	-2	-3	4	-39.6	53.1	-5.1	4	II	52.6	28.6	63.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1963	NM_017346	Huesken	Cacnb1	782	UUCUCGAGCUGGGCUAAGGcc	CCUUAGCCCAGCUCGAGAA	-3.3	-0.9	0.4	1	3	4	-42	62.5	0	5	Ib	57.9	64.1	76.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1964	NM_017346	Huesken	Cacnb1	782	GACACGUCGGAGUCUGACGgc	CGUCAGACUCCGACGUGUC	-2.4	-2.4	-3.2	2	1	3	-41.9	57.9	2.3	1	II	63.2	49.6	77	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1965	NM_017346	Huesken	Cacnb1	782	CGGAGUCUGACGGCCGGCUcg	AGCCGGCCGUCAGACUCCG	-2.1	-2.4	0	0	-2	8	-47.2	30.7	-8.3	1	III	73.7	35.2	44.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1966	NM_017346	Huesken	Cacnb1	782	GGCUCGUGUAGGACUCUGCug	GCAGAGUCCUACACGAGCC	-3.4	-3.3	1.2	1	0	3	-43.5	54.7	4.8	3	II	63.2	44.3	79.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1967	NM_017346	Huesken	Cacnb1	782	GACCUUUUGAACCGCCCUUuc	AAGGGCGGUUCAAAAGGUC	-0.9	-2.4	0	2	-1	6	-39	58.2	-1.6	4	II	52.6	42.2	57.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1968	NM_017346	Huesken	Cacnb1	782	CUGUAUUUGCCCUGUGGGCug	GCCCACAGGGCAAAUACAG	-3.4	-2.1	-6	0	1	4	-41.2	58.8	3.4	2	II	57.9	50.7	75.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1969	NM_017346	Huesken	Cacnb1	782	AGACCUCCAUAGGGAUCUCuu	GAGAUCCCUAUGGAGGUCU	-2.4	-2.1	-2.7	1	3	3	-41.4	48.8	9.7	2	II	52.6	45.3	71.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1970	NM_012864	Huesken	Mmp7	4316	UGUCGUCCUUUGUAAGACUga	AGUCUUACAAAGGACGACA	-2.1	-2.1	-4.1	1	1	2	-36.2	64.7	-4	3	II	42.1	57.1	98.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1971	NM_012864	Huesken	Mmp7	4316	UCCUUUGUAAGACUGAAGUcu	ACUUCAGUCUUACAAAGGA	-2.2	-2.4	0.3	0	1	2	-34.6	85	0.8	7	II	36.8	62	111.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1972	NM_012864	Huesken	Mmp7	4316	CCUUUGUAAGACUGAAGUCuu	GACUUCAGUCUUACAAAGG	-2.4	-3.3	-1.9	-1	0	2	-34.6	66.1	5.4	3	II	42.1	47.4	92.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1973	NM_012864	Huesken	Mmp7	4316	AAGACUGAAGUCUUCUGAAug	UUCAGAAGACUUCAGUCUU	-0.9	-0.9	-3.6	1	1	1	-34.7	73.8	-3.8	4	II	36.8	51.2	92.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1974	NM_012864	Huesken	Mmp7	4316	ACUGAAGUCUUCUGAAUGAuc	UCAUUCAGAAGACUUCAGU	-2.4	-2.2	1	1	0	1	-34.9	57.4	-3.4	6	II	36.8	43.6	54.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1975	NM_012864	Huesken	Mmp7	4316	UGAAUGAUCUCCUUGAUAGgu	CUAUCAAGGAGAUCAUUCA	-2.1	-2.1	-0.2	1	3	2	-34.3	87.2	-2.4	8	Ia	36.8	69.5	115.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1976	NM_012864	Huesken	Mmp7	4316	UGAUCUCCUUGAUAGGUAGgg	CUACCUAUCAAGGAGAUCA	-2.1	-2.1	-1.9	0	2	2	-36.7	77.8	0	5	Ib	42.1	58.9	114.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1977	NM_012864	Huesken	Mmp7	4316	AUCUCCUUGAUAGGUAGGGua	CCCUACCUAUCAAGGAGAU	-3.3	-1.1	-1.9	2	4	3	-38.8	66.3	0	5	Ib	47.4	60.8	102.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1978	NM_012864	Huesken	Mmp7	4316	AUAGGUAGGGUACAUCACAga	UGUGAUGUACCCUACCUAU	-2.1	-1.1	1.1	2	1	3	-37.7	53.7	-1.4	5	II	42.1	60.6	107.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1979	NM_012864	Huesken	Mmp7	4316	AGGUAGGGUACAUCACAGAac	UCUGUGAUGUACCCUACCU	-2.4	-2.1	1.1	1	-1	3	-39.8	50.9	-3.4	6	II	47.4	46.1	55.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1980	NM_012864	Huesken	Mmp7	4316	GUAGGGUACAUCACAGAACug	GUUCUGUGAUGUACCCUAC	-2.2	-2.2	1.9	2	1	3	-37.5	53.2	2.5	4	II	47.4	55.1	104.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1981	NM_012864	Huesken	Mmp7	4316	UAGGGUACAUCACAGAACUgg	AGUUCUGUGAUGUACCCUA	-2.1	-1.3	2.2	0	1	3	-37.4	65.1	-3.9	5	II	42.1	67.1	116.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1982	NM_012864	Huesken	Mmp7	4316	UCACAGAACUGGGAACAGAag	UCUGUUCCCAGUUCUGUGA	-2.4	-2.4	-0.3	2	0	3	-39.3	44	-1.1	6	II	47.4	51.2	99.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1983	NM_012864	Huesken	Mmp7	4316	GAACUGGGAACAGAAGAGUga	ACUCUUCUGUUCCCAGUUC	-2.2	-2.4	-0.3	1	0	3	-38.1	42.2	-6.3	5	III	47.4	43.4	81.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1984	NM_012864	Huesken	Mmp7	4316	CAGAAGAGUGACCCAGACCca	GGUCUGGGUCACUCUUCUG	-3.3	-2.1	-1.4	-1	0	3	-41.7	50.3	10.5	2	II	57.9	56.5	73.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1985	NM_012864	Huesken	Mmp7	4316	AAGAGUGACCCAGACCCAGag	CUGGGUCUGGGUCACUCUU	-2.1	-0.9	-1.2	2	3	3	-42.6	43.6	-2.4	3	Ib	57.9	57.1	87.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1986	NM_012864	Huesken	Mmp7	4316	GUGACCCAGACCCAGAGAGug	CUCUCUGGGUCUGGGUCAC	-2.1	-2.2	-0.4	1	1	3	-44.1	46.1	-2.3	2	II	63.2	40.5	73.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1987	NM_012864	Huesken	Mmp7	4316	UGACCCAGACCCAGAGAGUgg	ACUCUCUGGGUCUGGGUCA	-2.2	-2.1	-1	0	1	3	-44.1	42.7	-11.3	3	II	57.9	49.6	61.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1988	NM_012864	Huesken	Mmp7	4316	CCCAGACCCAGAGAGUGGCca	GCCACUCUCUGGGUCUGGG	-3.4	-3.3	-1.5	0	0	3	-46.2	32.8	5.4	0	II	68.4	37.7	63.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1989	NM_012864	Huesken	Mmp7	4316	CCAGACCCAGAGAGUGGCCaa	GGCCACUCUCUGGGUCUGG	-3.3	-3.3	-4.3	-2	1	4	-46.2	30.1	7.8	0	II	68.4	39.4	41.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1990	NM_012864	Huesken	Mmp7	4316	GGCCAAGUUCAUGAGUGGCaa	GCCACUCAUGAACUUGGCC	-3.4	-3.3	-3.3	1	1	4	-41.4	43.5	15.4	0	II	57.9	36.8	62.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1991	NM_012864	Huesken	Mmp7	4316	UCAUGAGUGGCAACAAACAgg	UGUUUGUUGCCACUCAUGA	-2.1	-2.4	1.5	1	0	3	-36.6	53.8	-8.7	6	II	42.1	56.4	104.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1992	NM_012864	Huesken	Mmp7	4316	GAGUGGCAACAAACAGGAAgu	UUCCUGUUUGUUGCCACUC	-0.9	-2.4	1.5	0	-2	3	-37.6	41.6	-4	2	III	47.4	25.9	37.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1993	NM_012864	Huesken	Mmp7	4316	GGCAACAAACAGGAAGUUCac	GAACUUCCUGUUUGUUGCC	-2.4	-3.3	-1	0	-1	3	-36.4	52.7	12.4	3	II	47.4	40.7	53.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1994	NM_012864	Huesken	Mmp7	4316	GCAACAAACAGGAAGUUCAcg	UGAACUUCCUGUUUGUUGC	-2.1	-3.4	-1.2	1	-2	2	-35.2	59	-6.3	4	II	42.1	42.2	33.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1995	NM_012864	Huesken	Mmp7	4316	AAACAGGAAGUUCACGCCUga	AGGCGUGAACUUCCUGUUU	-2.1	-0.9	-0.5	2	3	4	-37.9	52.6	4.2	6	II	47.4	61	91.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1996	NM_012864	Huesken	Mmp7	4316	AGGAAGUUCACGCCUGAGUcc	ACUCAGGCGUGAACUUCCU	-2.2	-2.1	0.5	3	0	4	-40.6	60.2	-1.2	6	II	52.6	52.6	73.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1997	NM_012864	Huesken	Mmp7	4316	GAAGUUCACGCCUGAGUCCuc	GGACUCAGGCGUGAACUUC	-3.3	-2.4	0.5	0	2	4	-40.9	51.6	2.7	3	II	57.9	50.4	70.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1998	NM_012864	Huesken	Mmp7	4316	UUCACGCCUGAGUCCUCACca	GUGAGGACUCAGGCGUGAA	-2.2	-0.9	-0.6	0	3	4	-42.1	67.3	9.7	4	II	57.9	62.6	89.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si1999	NM_012864	Huesken	Mmp7	4316	UCACGCCUGAGUCCUCACCau	GGUGAGGACUCAGGCGUGA	-3.3	-2.4	-0.6	2	3	4	-44.5	57.9	12.8	4	II	63.2	62.5	94.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2000	NM_012864	Huesken	Mmp7	4316	CACGCCUGAGUCCUCACCAuc	UGGUGAGGACUCAGGCGUG	-2.1	-2.1	-0.6	0	-1	4	-44.2	48.6	-3.4	1	III	63.2	43.5	69.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2001	NM_012864	Huesken	Mmp7	4316	ACGCCUGAGUCCUCACCAUcc	AUGGUGAGGACUCAGGCGU	-1.1	-2.2	-0.6	3	0	4	-43.2	43.1	-8.7	2	II	57.9	38.1	65.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2002	NM_012864	Huesken	Mmp7	4316	CGCCUGAGUCCUCACCAUCcg	GAUGGUGAGGACUCAGGCG	-2.4	-2.4	-0.6	0	-2	4	-43.4	42.7	8.5	1	II	63.2	43.3	61.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2003	NM_012864	Huesken	Mmp7	4316	GAGUCCUCACCAUCCGUCCag	GGACGGAUGGUGAGGACUC	-3.3	-2.4	0.8	1	1	3	-43.7	62.2	5.1	3	II	63.2	51.8	81.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2004	NM_012864	Huesken	Mmp7	4316	AGUCCUCACCAUCCGUCCAgu	UGGACGGAUGGUGAGGACU	-2.1	-2.1	0.9	3	1	3	-43.4	58.5	4	2	II	57.9	42.9	52.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2005	NM_012864	Huesken	Mmp7	4316	UCCUCACCAUCCGUCCAGUac	ACUGGACGGAUGGUGAGGA	-2.2	-2.4	0.3	0	1	3	-43.4	52.6	-6.3	3	II	57.9	44.4	46.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2006	NM_012864	Huesken	Mmp7	4316	CCUCACCAUCCGUCCAGUAcu	UACUGGACGGAUGGUGAGG	-1.3	-3.3	-0.8	0	-3	3	-42.3	39.7	-5.4	1	III	57.9	26.6	32.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2007	NM_012864	Huesken	Mmp7	4316	UCACCAUCCGUCCAGUACUca	AGUACUGGACGGAUGGUGA	-2.1	-2.4	-1	3	1	3	-41.2	58.6	-3.9	4	II	52.6	61.3	90.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2008	NM_012864	Huesken	Mmp7	4316	ACCAUCCGUCCAGUACUCAuc	UGAGUACUGGACGGAUGGU	-2.1	-2.2	0.7	1	0	3	-41.2	49.4	-3.8	5	II	52.6	40.6	63.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2009	NM_012864	Huesken	Mmp7	4316	CCAUCCGUCCAGUACUCAUcc	AUGAGUACUGGACGGAUGG	-1.1	-3.3	0.7	0	-2	3	-40.1	58.2	-3.7	3	III	52.6	35.8	41.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2010	NM_012864	Huesken	Mmp7	4316	CAUCCGUCCAGUACUCAUCcu	GAUGAGUACUGGACGGAUG	-2.4	-2.1	0.7	2	0	3	-39.2	68.7	5.8	4	II	52.6	56.8	87.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2011	NM_012864	Huesken	Mmp7	4316	CCGUCCAGUACUCAUCCUUgu	AAGGAUGAGUACUGGACGG	-0.9	-3.3	0.6	-1	-3	3	-39.9	56.1	-2.9	3	III	52.6	38.8	75.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2012	NM_012864	Huesken	Mmp7	4316	CAUCCUUGUCAAAGUGAGCau	GCUCACUUUGACAAGGAUG	-3.4	-2.1	0.7	0	1	2	-36.8	62.2	5.1	3	II	47.4	52.8	101.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2013	NM_012864	Huesken	Mmp7	4316	UCAAAGUGAGCAUCUCCGCcg	GCGGAGAUGCUCACUUUGA	-3.4	-2.4	-1.8	1	4	4	-39.8	60.7	7.8	6	Ia	52.6	69.3	108.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2014	NM_012864	Huesken	Mmp7	4316	AAAGUGAGCAUCUCCGCCGag	CGGCGGAGAUGCUCACUUU	-2.4	-0.9	-2.1	2	5	6	-41	58.7	-2.4	5	Ia	57.9	74	118.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2015	NM_012864	Huesken	Mmp7	4316	GUGAGCAUCUCCGCCGAGGcc	CCUCGGCGGAGAUGCUCAC	-3.3	-2.2	-2.1	1	1	6	-44.9	41.5	-4.7	1	II	68.4	46.3	72.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2016	NM_012864	Huesken	Mmp7	4316	UGAGCAUCUCCGCCGAGGCcu	GCCUCGGCGGAGAUGCUCA	-3.4	-2.1	-2.4	1	4	6	-46.1	49.8	7.5	3	Ib	68.4	61.7	90.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2017	NM_012864	Huesken	Mmp7	4316	AGCAUCUCCGCCGAGGCCUgg	AGGCCUCGGCGGAGAUGCU	-2.1	-2.1	-2.4	3	1	6	-47	45.2	-8.7	3	II	68.4	50.3	68.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2018	NM_012864	Huesken	Mmp7	4316	AUCUCCGCCGAGGCCUGGCcc	GCCAGGCCUCGGCGGAGAU	-3.4	-1.1	-4	2	4	6	-48.2	48.9	9.7	3	II	73.7	55.3	36.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2019	NM_012864	Huesken	Mmp7	4316	UCCGCCGAGGCCUGGCCCCgg	GGGGCCAGGCCUCGGCGGA	-3.3	-2.4	-6.3	0	3	6	-52.5	29.8	2.7	2	II	84.2	53.9	50.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2020	NM_012864	Huesken	Mmp7	4316	GCCGAGGCCUGGCCCCGGUgc	ACCGGGGCCAGGCCUCGGC	-2.2	-3.4	-11.2	1	-1	8	-52.3	15.1	1.1	0	III	84.2	23.7	11	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2021	NM_012864	Huesken	Mmp7	4316	CUGGCCCCGGUGCAAAGGCau	GCCUUUGCACCGGGGCCAG	-3.4	-2.1	-4.8	0	1	8	-46.9	31.8	7.8	-1	II	73.7	44.9	32.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2022	NM_012864	Huesken	Mmp7	4316	UGGCCCCGGUGCAAAGGCAug	UGCCUUUGCACCGGGGCCA	-2.1	-2.1	-4.8	2	0	8	-46.9	33.2	-8.7	1	II	68.4	41	59.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2023	NM_012864	Huesken	Mmp7	4316	AAAGGCAUGGCCUAGAGUGuu	CACUCUAGGCCAUGCCUUU	-2.1	-0.9	-2.9	2	4	4	-40.2	56	-4.7	4	Ib	52.6	63.3	100.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2024	NM_012864	Huesken	Mmp7	4316	GCAUGGCCUAGAGUGUUUCcu	GAAACACUCUAGGCCAUGC	-2.4	-3.4	1.5	0	1	4	-39.4	53.3	9.7	4	II	52.6	42.2	53.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2025	NM_012864	Huesken	Mmp7	4316	CAUGGCCUAGAGUGUUUCCug	GGAAACACUCUAGGCCAUG	-3.3	-2.1	0.4	-2	1	4	-39.3	62.2	7.7	3	II	52.6	48.2	89.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2026	NM_012864	Huesken	Mmp7	4316	GGCCUAGAGUGUUUCCUGGcc	CCAGGAAACACUCUAGGCC	-3.3	-3.3	0.1	2	1	4	-41.5	54.4	5.4	2	II	57.9	37.8	33.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2027	NM_012864	Huesken	Mmp7	4316	CCUAGAGUGUUUCCUGGCCca	GGCCAGGAAACACUCUAGG	-3.3	-3.3	0.1	0	1	4	-41.5	50.5	10.8	2	II	57.9	45	54.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2028	NM_012864	Huesken	Mmp7	4316	CUAGAGUGUUUCCUGGCCCau	GGGCCAGGAAACACUCUAG	-3.3	-2.1	0.1	0	3	5	-41.5	52.5	10.5	3	II	57.9	60.4	83.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2029	NM_012864	Huesken	Mmp7	4316	GUUUCCUGGCCCAUCAAAUgg	AUUUGAUGGGCCAGGAAAC	-1.1	-2.2	0.1	2	0	5	-37.8	66.6	-8.9	4	II	47.4	41.5	10.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2030	NM_012864	Huesken	Mmp7	4316	UGGCCCAUCAAAUGGGAAGuu	CUUCCCAUUUGAUGGGCCA	-2.1	-2.1	-4.1	1	1	5	-40.1	64.6	2.3	4	II	52.6	57.3	85	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2031	NM_012864	Huesken	Mmp7	4316	GCCCAUCAAAUGGGAAGUUgu	AACUUCCCAUUUGAUGGGC	-0.9	-3.4	-4.1	0	-2	4	-37.8	36.4	-1.3	1	III	47.4	25.8	0.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2032	NM_012864	Huesken	Mmp7	4316	GUUGUCUCCGUGAUCUCCCcu	GGGAGAUCACGGAGACAAC	-3.3	-2.2	0.8	2	4	3	-41.3	63.5	7.1	3	II	57.9	61.5	94.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2033	NM_012864	Huesken	Mmp7	4316	UGUCUCCGUGAUCUCCCCUug	AGGGGAGAUCACGGAGACA	-2.1	-2.1	-0.7	1	3	4	-43.6	64.4	6.1	4	II	57.9	57.1	101.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2034	NM_012864	Huesken	Mmp7	4316	CUCCGUGAUCUCCCCUUGCga	GCAAGGGGAGAUCACGGAG	-3.4	-2.1	-0.7	0	0	4	-43.3	47.7	5.5	2	II	63.2	51.1	90.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2035	NM_012864	Huesken	Mmp7	4316	UCCGUGAUCUCCCCUUGCGaa	CGCAAGGGGAGAUCACGGA	-2.4	-2.4	-0.7	1	3	4	-43.6	55.4	-2.3	4	II	63.2	60.9	97.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2036	NM_012864	Huesken	Mmp7	4316	CGUGAUCUCCCCUUGCGAAgc	UUCGCAAGGGGAGAUCACG	-0.9	-2.4	-1.9	0	-2	4	-41.2	49.4	-5.4	0	III	57.9	28	26.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2037	NM_012864	Huesken	Mmp7	4316	UGAUCUCCCCUUGCGAAGCca	GCUUCGCAAGGGGAGAUCA	-3.4	-2.1	-5	1	3	4	-42.1	65.5	10.4	3	Ib	57.9	64.4	105.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2038	NM_012864	Huesken	Mmp7	4316	CCUUGCGAAGCCAAUUAUGau	CAUAAUUGGCUUCGCAAGG	-2.1	-3.3	-1	-1	-1	3	-35.8	53.4	-9.3	2	II	47.4	43.4	63.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2039	NM_012864	Huesken	Mmp7	4316	CUUGCGAAGCCAAUUAUGAug	UCAUAAUUGGCUUCGCAAG	-2.4	-2.1	0.4	0	-1	3	-34.9	59.1	-8.4	4	III	42.1	46.2	75.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2040	NM_012864	Huesken	Mmp7	4316	GAAGCCAAUUAUGAUGUCUgc	AGACAUCAUAAUUGGCUUC	-2.1	-2.4	0.6	0	0	3	-33.9	68.6	1.4	4	III	36.8	47.9	99.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2041	NM_012864	Huesken	Mmp7	4316	AAGCCAAUUAUGAUGUCUGca	CAGACAUCAUAAUUGGCUU	-2.1	-0.9	0.6	2	4	3	-33.6	74.7	-0.3	5	Ib	36.8	66.6	102.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2042	NM_012864	Huesken	Mmp7	4316	AGCCAAUUAUGAUGUCUGCag	GCAGACAUCAUAAUUGGCU	-3.4	-2.1	0.6	2	2	3	-36.1	65.6	10.1	5	Ib	42.1	50.9	95.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2043	NM_012864	Huesken	Mmp7	4316	CCCCAACUAACCCUCUUGAag	UCAAGAGGGUUAGUUGGGG	-2.4	-3.3	1.4	0	-3	4	-40.2	49.7	-3	3	III	52.6	29.4	40.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2044	NM_012864	Huesken	Mmp7	4316	AGUGGGAUUUGCAUACUCCac	GGAGUAUGCAAAUCCCACU	-3.3	-2.1	-1.2	1	4	3	-38.3	63.7	4.8	4	II	47.4	63.2	96.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2045	NM_012864	Huesken	Mmp7	4316	CUGGAAUGCCACUUAGGACug	GUCCUAAGUGGCAUUCCAG	-2.2	-2.1	-1.1	1	1	3	-39.3	57.7	10.4	2	II	52.6	48.3	109.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2046	NM_012864	Huesken	Mmp7	4316	GAAUGCCACUUAGGACUGUuu	ACAGUCCUAAGUGGCAUUC	-2.2	-2.4	-1.3	0	0	3	-38.2	43.1	-4	3	III	47.4	34.4	65.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2047	NM_012864	Huesken	Mmp7	4316	UGCCACUUAGGACUGUUUGgc	CAAACAGUCCUAAGUGGCA	-2.1	-2.1	-1.3	2	2	3	-37.7	74.6	7.4	5	II	47.4	60.7	81.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2048	NM_012864	Huesken	Mmp7	4316	UUAGGACUGUUUGGCAUUAgu	UAAUGCCAAACAGUCCUAA	-1.3	-0.9	1.8	0	0	3	-34.6	58.9	-0.7	6	II	36.8	52.8	105	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2049	NM_012864	Huesken	Mmp7	4316	UGGCAUUAGUGAGAAUUCUgc	AGAAUUCUCACUAAUGCCA	-2.1	-2.1	0.8	3	0	3	-34.9	65.6	-1.3	6	II	36.8	59.4	106.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2050	NM_012864	Huesken	Mmp7	4316	GCAUUAGUGAGAAUUCUGCaa	GCAGAAUUCUCACUAAUGC	-3.4	-3.4	0.2	1	2	2	-35	68	7.1	6	II	42.1	51.4	94.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2051	NM_012864	Huesken	Mmp7	4316	CAUUAGUGAGAAUUCUGCAac	UGCAGAAUUCUCACUAAUG	-2.1	-2.1	0.8	0	0	2	-33.7	75.1	1.6	5	II	36.8	49.7	72.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2052	NM_012864	Huesken	Mmp7	4316	UUAGUGAGAAUUCUGCAACau	GUUGCAGAAUUCUCACUAA	-2.2	-0.9	0.8	0	4	2	-33.6	82.9	15.2	8	Ia	36.8	77.6	122.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2053	NM_012864	Huesken	Mmp7	4316	AGUGAGAAUUCUGCAACAUcu	AUGUUGCAGAAUUCUCACU	-1.1	-2.1	0.8	1	1	2	-34.6	57.2	-0.6	5	II	36.8	45.9	80.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2054	NM_012864	Huesken	Mmp7	4316	GUGAGAAUUCUGCAACAUCug	GAUGUUGCAGAAUUCUCAC	-2.4	-2.2	0.8	0	0	2	-34.9	57.4	9.8	3	II	42.1	52.1	119.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2055	NM_012864	Huesken	Mmp7	4316	GAAUUCUGCAACAUCUGGCac	GCCAGAUGUUGCAGAAUUC	-3.4	-2.4	-0.3	2	4	3	-37	77	7.1	6	II	47.4	61.1	100.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2056	NM_012864	Huesken	Mmp7	4316	UUCUGCAACAUCUGGCACUcc	AGUGCCAGAUGUUGCAGAA	-2.1	-0.9	-1.2	0	1	3	-39	69.4	-6.3	6	II	47.4	66.4	115	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2057	NM_012864	Huesken	Mmp7	4316	UCUGCAACAUCUGGCACUCca	GAGUGCCAGAUGUUGCAGA	-2.4	-2.4	-1.2	-1	3	3	-40.5	54.9	8.1	4	Ib	52.6	57.3	104	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2058	NM_012864	Huesken	Mmp7	4316	UGCAACAUCUGGCACUCCAca	UGGAGUGCCAGAUGUUGCA	-2.1	-2.1	-1.4	1	0	3	-41.4	61.8	1.4	5	II	52.6	54.3	91.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2059	NM_012864	Huesken	Mmp7	4316	AACAUCUGGCACUCCACACcu	GUGUGGAGUGCCAGAUGUU	-2.2	-0.9	-1.6	3	3	3	-40.3	71.6	7.4	5	Ia	52.6	61.1	93.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2060	NM_012864	Huesken	Mmp7	4316	UCUGGCACUCCACACCUGGgc	CCAGGUGUGGAGUGCCAGA	-3.3	-2.4	-2.2	0	3	3	-44.8	48.4	0.1	4	II	63.2	62.8	85.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2061	NM_012864	Huesken	Mmp7	4316	CUGGCACUCCACACCUGGGcu	CCCAGGUGUGGAGUGCCAG	-3.3	-2.1	-2.2	0	1	3	-45.7	42.3	-4.6	-1	II	68.4	44.8	60.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2062	NM_012864	Huesken	Mmp7	4316	GCACUCCACACCUGGGCUUcu	AAGCCCAGGUGUGGAGUGC	-0.9	-3.4	-1.4	1	-1	4	-44.6	37.2	-8.7	1	III	63.2	29	47.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2063	NM_012864	Huesken	Mmp7	4316	CACUCCACACCUGGGCUUCug	GAAGCCCAGGUGUGGAGUG	-2.4	-2.1	-1.4	-2	-1	4	-43.6	51.8	6.1	2	II	63.2	41.9	94	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2064	NM_012864	Huesken	Mmp7	4316	CCACACCUGGGCUUCUGCAuu	UGCAGAAGCCCAGGUGUGG	-2.1	-3.3	-0.3	1	-1	4	-44.5	42.3	-3.1	1	III	63.2	32.5	61.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2065	NM_012864	Huesken	Mmp7	4316	ACACCUGGGCUUCUGCAUUau	AAUGCAGAAGCCCAGGUGU	-0.9	-2.2	-0.3	3	1	4	-41.1	54.4	3.8	4	II	52.6	47.2	86.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2066	NM_012864	Huesken	Mmp7	4316	CUUCUGCAUUAUCUCCAUGac	CAUGGAGAUAAUGCAGAAG	-2.1	-2.1	0.9	0	1	2	-35	71.3	8.1	3	II	42.1	50.9	97.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2067	NM_012864	Huesken	Mmp7	4316	UCUGCAUUAUCUCCAUGACac	GUCAUGGAGAUAAUGCAGA	-2.2	-2.4	0.1	1	3	2	-36.6	75.7	10.5	5	Ib	42.1	56.5	115	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2068	NM_012864	Huesken	Mmp7	4316	UGCAUUAUCUCCAUGACACgg	GUGUCAUGGAGAUAAUGCA	-2.2	-2.1	-1.2	1	3	2	-36.4	81.5	2.4	5	Ia	42.1	68	112.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2069	NM_012864	Huesken	Mmp7	4316	GGACAGCUUUCCAGUCUCCgg	GGAGACUGGAAAGCUGUCC	-3.3	-3.3	-1.2	0	1	2	-41.7	39.3	2.8	1	II	57.9	40.4	88.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2070	NM_012864	Huesken	Mmp7	4316	GCUUUCCAGUCUCCGGCAAac	UUGCCGGAGACUGGAAAGC	-0.9	-3.4	-2.4	1	-2	5	-41.7	47.4	-0.7	1	III	57.9	18.8	1.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2071	NM_012864	Huesken	Mmp7	4316	UUCCAGUCUCCGGCAAACCga	GGUUUGCCGGAGACUGGAA	-3.3	-0.9	-2.4	2	3	5	-41.7	55.5	7.8	4	Ib	57.9	70.4	102.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2072	NM_012864	Huesken	Mmp7	4316	UCCGGCAAACCGAAGAACUuc	AGUUCUUCGGUUUGCCGGA	-2.1	-2.4	-4.7	0	0	5	-39.6	48.7	-8.9	3	II	52.6	56.1	53.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2073	NM_012864	Huesken	Mmp7	4316	CCGGCAAACCGAAGAACUUcu	AAGUUCUUCGGUUUGCCGG	-0.9	-3.3	-3.7	-1	-3	5	-38.1	33.2	-8.6	0	III	52.6	32	1.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2074	NM_012864	Huesken	Mmp7	4316	GAACUUCUGCAUUUCCCUCag	GAGGGAAAUGCAGAAGUUC	-2.4	-2.4	1.8	2	3	3	-37	73.4	15.1	3	II	47.4	53.1	97.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2075	NM_012864	Huesken	Mmp7	4316	AACUUCUGCAUUUCCCUCAgu	UGAGGGAAAUGCAGAAGUU	-2.1	-0.9	1.8	3	0	3	-36.7	85.2	-0.7	8	II	42.1	62	103.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2076	NM_012864	Huesken	Mmp7	4316	ACUUCUGCAUUUCCCUCAGuu	CUGAGGGAAAUGCAGAAGU	-2.1	-2.2	-0.2	1	3	3	-37.9	71.2	5.4	5	Ib	47.4	51.6	83.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2077	NM_012864	Huesken	Mmp7	4316	UCUGCAUUUCCCUCAGUUUgu	AAACUGAGGGAAAUGCAGA	-0.9	-2.4	-0.7	1	0	3	-36.7	73.4	-8.7	6	II	42.1	55.8	116.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2078	NM_012864	Huesken	Mmp7	4316	CAUUUCCCUCAGUUUGUCCac	GGACAAACUGAGGGAAAUG	-3.3	-2.1	2.2	-1	2	3	-36.7	83.1	10	5	II	47.4	55.7	83	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2079	NM_012864	Huesken	Mmp7	4316	CCCUCAGUUUGUCCACUGCac	GCAGUGGACAAACUGAGGG	-3.4	-3.3	-3.7	-1	-1	3	-41.2	57.1	10.8	1	II	57.9	39.2	32.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2080	NM_012864	Huesken	Mmp7	4316	UCAGUUUGUCCACUGCACUgg	AGUGCAGUGGACAAACUGA	-2.1	-2.4	-3.7	1	2	2	-38.9	67.3	-1.6	5	II	47.4	65.1	110.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2081	NM_012864	Huesken	Mmp7	4316	CAGUUUGUCCACUGCACUGgu	CAGUGCAGUGGACAAACUG	-2.1	-2.1	-2.6	-1	0	2	-38.6	65.1	-2.1	3	II	52.6	52.5	100	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2082	NM_012864	Huesken	Mmp7	4316	CCACUGCACUGGUGGCCUUcu	AAGGCCACCAGUGCAGUGG	-0.9	-3.3	-2.9	0	-2	4	-44.3	32.7	-6	0	III	63.2	27.5	22.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2083	NM_012864	Huesken	Mmp7	4316	GCACUGGUGGCCUUCUUUGuu	CAAAGAAGGCCACCAGUGC	-2.1	-3.4	0.5	2	1	4	-40.9	51.6	-2.4	3	II	57.9	44	38.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2084	NM_012864	Huesken	Mmp7	4316	ACUGGUGGCCUUCUUUGUUuu	AACAAAGAAGGCCACCAGU	-0.9	-2.2	0.5	2	2	4	-38.5	60.9	3.8	4	II	47.4	47.1	87.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2085	NM_012864	Huesken	Mmp7	4316	UGGUGGCCUUCUUUGUUUUag	AAAACAAAGAAGGCCACCA	-0.9	-2.1	0.5	0	0	4	-36	76.1	-1	5	II	42.1	51.2	45.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2086	NM_012864	Huesken	Mmp7	4316	UGGCCUUCUUUGUUUUAGAgu	UCUAAAACAAAGAAGGCCA	-2.4	-2.1	0.6	2	0	4	-34.2	81.4	-1.4	4	II	36.8	56.2	95.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2087	NM_012864	Huesken	Mmp7	4316	CUUCUUUGUUUUAGAGUCGug	CGACUCUAAAACAAAGAAG	-2.4	-2.1	0.5	0	1	2	-31.2	73.8	-1.6	5	II	36.8	59.2	113.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2088	NM_012864	Huesken	Mmp7	4316	CUUUGUUUUAGAGUCGUGAag	UCACGACUCUAAAACAAAG	-2.4	-2.1	3.2	0	-1	2	-32.5	79.5	-3.4	6	II	36.8	52.2	108	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2089	NM_012864	Huesken	Mmp7	4316	UGUUUUAGAGUCGUGAAGGua	CCUUCACGACUCUAAAACA	-3.3	-2.1	4.8	0	4	2	-34.9	78.4	0	7	Ia	42.1	71.2	126.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2090	NM_012864	Huesken	Mmp7	4316	UUUUAGAGUCGUGAAGGUAaa	UACCUUCACGACUCUAAAA	-1.3	-0.9	4.8	0	1	2	-34.1	68.2	2	8	II	36.8	58.1	110	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2091	NM_012864	Huesken	Mmp7	4316	UUUAGAGUCGUGAAGGUAAaa	UUACCUUCACGACUCUAAA	-0.9	-0.9	4	1	0	2	-34.1	68.3	-6.3	8	II	36.8	59.6	130.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2092	NM_012864	Huesken	Mmp7	4316	UUAGAGUCGUGAAGGUAAAau	UUUACCUUCACGACUCUAA	-0.9	-0.9	3.8	1	0	2	-34.1	59.7	-3.6	7	II	36.8	54.1	110	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2093	NM_012864	Huesken	Mmp7	4316	GAGUCGUGAAGGUAAAAUUuc	AAUUUUACCUUCACGACUC	-0.9	-2.4	3.8	1	-2	2	-32.7	71.9	-4	4	III	36.8	45.4	76.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2094	NM_012864	Huesken	Mmp7	4316	AGUCGUGAAGGUAAAAUUUcc	AAAUUUUACCUUCACGACU	-0.9	-2.1	3.8	2	0	2	-31.2	68.4	-3.5	5	II	31.6	52.2	74.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2095	NM_012864	Huesken	Mmp7	4316	UCGUGAAGGUAAAAUUUCCua	GGAAAUUUUACCUUCACGA	-3.3	-2.4	0	1	2	2	-32.6	69.8	7.1	6	Ib	36.8	67.3	126.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2096	NM_012864	Huesken	Mmp7	4316	GUGAAGGUAAAAUUUCCUAag	UAGGAAAUUUUACCUUCAC	-1.3	-2.2	-0.7	0	-1	2	-31.2	73.1	3.6	5	III	31.6	46.2	86.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2097	NM_012864	Huesken	Mmp7	4316	AGGUAAAAUUUCCUAAGAUaa	AUCUUAGGAAAUUUUACCU	-1.1	-2.1	-0.6	1	1	2	-30.1	70.4	1.1	5	II	26.3	47	79.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2098	NM_012864	Huesken	Mmp7	4316	GGUAAAAUUUCCUAAGAUAau	UAUCUUAGGAAAUUUUACC	-1.3	-3.3	0.8	0	-3	2	-29.3	67.8	-5.8	4	II	26.3	35.7	28.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2099	NM_012864	Huesken	Mmp7	4316	AAAUUUCCUAAGAUAAUUCug	GAAUUAUCUUAGGAAAUUU	-2.4	-0.9	0.8	1	4	2	-26.9	99.6	9.4	8	Ia	21.1	74.5	116.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2100	NM_012864	Huesken	Mmp7	4316	UUUCCUAAGAUAAUUCUGCgc	GCAGAAUUAUCUUAGGAAA	-3.4	-0.9	0.8	2	5	2	-31.6	89.9	7.1	8	Ia	31.6	83.7	131.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2101	NM_012864	Huesken	Mmp7	4316	CCUAAGAUAAUUCUGCGCCug	GGCGCAGAAUUAUCUUAGG	-3.3	-3.3	0.9	-1	2	5	-36.5	69.1	15.9	5	II	47.4	58.3	134.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2102	NM_012864	Huesken	Mmp7	4316	CUAAGAUAAUUCUGCGCCUgu	AGGCGCAGAAUUAUCUUAG	-2.1	-2.1	0.9	-1	0	5	-35.3	57.9	-8.3	4	II	42.1	50.1	66	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2103	NM_012864	Huesken	Mmp7	4316	AAGAUAAUUCUGCGCCUGUuc	ACAGGCGCAGAAUUAUCUU	-2.2	-0.9	0.9	0	1	5	-36.2	64.2	-1.6	7	II	42.1	56	75.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2104	NM_012864	Huesken	Mmp7	4316	AUAAUUCUGCGCCUGUUCCca	GGAACAGGCGCAGAAUUAU	-3.3	-1.1	3.6	2	5	5	-37.4	81.9	9.8	7	Ia	47.4	69.6	104.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2105	NM_012864	Huesken	Mmp7	4316	UAAUUCUGCGCCUGUUCCCac	GGGAACAGGCGCAGAAUUA	-3.3	-1.3	3.6	1	5	5	-39.6	76	2.7	8	Ia	52.6	74.9	93.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2106	NM_012864	Huesken	Mmp7	4316	AAUUCUGCGCCUGUUCCCAcu	UGGGAACAGGCGCAGAAUU	-2.1	-0.9	-0.7	2	3	5	-40.4	66	-0.8	6	II	52.6	52.9	46.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2107	NM_012864	Huesken	Mmp7	4316	CUGUUCCCACUGAAGUGCGgu	CGCACUUCAGUGGGAACAG	-2.4	-2.1	-1.9	-1	1	3	-40.3	53.3	-4.6	1	II	57.9	50.5	91.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2108	NM_012864	Huesken	Mmp7	4316	UGUUCCCACUGAAGUGCGGuc	CCGCACUUCAGUGGGAACA	-3.3	-2.1	-4.6	0	4	4	-41.5	56.1	-2.6	3	II	57.9	52.9	91.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2109	NM_012864	Huesken	Mmp7	4316	CACUGAAGUGCGGUCACUUcu	AAGUGACCGCACUUCAGUG	-0.9	-2.1	-2.6	-1	-1	4	-39.2	47.5	-6	2	II	52.6	40.9	79.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2110	NM_012864	Huesken	Mmp7	4316	ACUGAAGUGCGGUCACUUCuc	GAAGUGACCGCACUUCAGU	-2.4	-2.2	-7.1	2	2	4	-39.5	54.7	10.1	5	Ib	52.6	51.5	40.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2111	NM_012864	Huesken	Mmp7	4316	CUGAAGUGCGGUCACUUCUcc	AGAAGUGACCGCACUUCAG	-2.1	-2.1	-7.3	1	-2	4	-39.4	61.7	2.2	4	II	52.6	52.5	32.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2112	NM_012864	Huesken	Mmp7	4316	AGUGCGGUCACUUCUCCGGcu	CCGGAGAAGUGACCGCACU	-3.3	-2.1	-1.9	2	4	4	-43.3	58.9	0.7	3	II	63.2	57.9	94.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2113	NM_012864	Huesken	Mmp7	4316	GGUCACUUCUCCGGCUUCCug	GGAAGCCGGAGAAGUGACC	-3.3	-3.3	0.7	2	1	5	-43.2	47	5.4	3	II	63.2	41.9	83	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2114	NM_012864	Huesken	Mmp7	4316	CACUUCUCCGGCUUCCUGGga	CCAGGAAGCCGGAGAAGUG	-3.3	-2.1	-0.3	1	1	5	-42.8	69.7	-2	4	II	63.2	54.4	72.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2115	NM_012864	Huesken	Mmp7	4316	ACUUCUCCGGCUUCCUGGGac	CCCAGGAAGCCGGAGAAGU	-3.3	-2.2	-0.6	2	4	5	-44	61.1	0.7	3	Ib	63.2	50.7	64.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2116	NM_012864	Huesken	Mmp7	4316	GGCUUCCUGGGACAGUGGCag	GCCACUGUCCCAGGAAGCC	-3.4	-3.3	-0.3	1	1	3	-46	44	9.8	1	II	68.4	34.4	66	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2117	NM_012864	Huesken	Mmp7	4316	UUCCUGGGACAGUGGCAGGgc	CCUGCCACUGUCCCAGGAA	-3.3	-0.9	-2.9	0	3	3	-44.7	50.8	2.3	4	II	63.2	60	67.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2118	NM_012864	Huesken	Mmp7	4316	UGGGACAGUGGCAGGGCCAgg	UGGCCCUGCCACUGUCCCA	-2.1	-2.1	-3.1	-1	0	5	-48.1	32.3	-6	3	II	68.4	48.6	93.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2119	NM_012864	Huesken	Mmp7	4316	AGUGGCAGGGCCAGGCAGCcu	GCUGCCUGGCCCUGCCACU	-3.4	-2.1	-3.2	1	3	5	-49.1	28	0.1	2	II	73.7	49.7	64.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2120	NM_025237	Huesken	SOST	50964	GGCCGAGCGGCCUCGGUCCcg	GGACCGAGGCCGCUCGGCC	-3.3	-3.3	-6.9	2	0	6	-50.8	24	2.7	0	II	84.2	35.4	50.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2121	NM_025237	Huesken	SOST	50964	GGCCUCGGUCCCGAAGUCCuu	GGACUUCGGGACCGAGGCC	-3.3	-3.3	-5.7	1	0	4	-47.1	29.4	2.7	1	II	73.7	37.7	53.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2122	NM_025237	Huesken	SOST	50964	CCUCGGUCCCGAAGUCCUUga	AAGGACUUCGGGACCGAGG	-0.9	-3.3	-5.3	1	-2	4	-43.4	31.3	-8.6	0	III	63.2	32.2	58.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2123	NM_025237	Huesken	SOST	50964	UCGGUCCCGAAGUCCUUGAgc	UCAAGGACUUCGGGACCGA	-2.4	-2.4	-1.8	1	-1	4	-42.5	59.2	-1.5	4	II	57.9	49.8	61.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2124	NM_025237	Huesken	SOST	50964	CGGUCCCGAAGUCCUUGAGcu	CUCAAGGACUUCGGGACCG	-2.1	-2.4	0.1	-1	-1	4	-42.2	54.8	3.4	1	II	63.2	31.5	62.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2125	NM_025237	Huesken	SOST	50964	GUCCCGAAGUCCUUGAGCUcc	AGCUCAAGGACUUCGGGAC	-2.1	-2.2	0	2	1	4	-42	50.7	-6.3	1	III	57.9	41.1	64.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2126	NM_025237	Huesken	SOST	50964	CCCGAAGUCCUUGAGCUCCga	GGAGCUCAAGGACUUCGGG	-3.3	-3.3	-2.1	0	-1	4	-43.1	39.4	11.1	1	II	63.2	45.6	75	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2127	NM_025237	Huesken	SOST	50964	CGAAGUCCUUGAGCUCCGAcu	UCGGAGCUCAAGGACUUCG	-2.4	-2.4	-2.9	-1	-2	3	-41.3	38	-5.8	0	III	57.9	31.1	61.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2128	NM_025237	Huesken	SOST	50964	GAGCUCCGACUGGUUGUGGaa	CCACAACCAGUCGGAGCUC	-3.3	-2.4	-0.8	1	2	3	-43	45.1	2.3	2	II	63.2	39.8	80.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2129	NM_025237	Huesken	SOST	50964	AGCGGGUGAGGCGCUUGCAcu	UGCAAGCGCCUCACCCGCU	-2.1	-2.1	-2.5	2	0	5	-46.4	32.4	-6.1	1	II	68.4	38.3	66.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2130	NM_025237	Huesken	SOST	50964	GGUGAGGCGCUUGCACUUGca	CAAGUGCAAGCGCCUCACC	-2.1	-3.3	-1.7	1	0	5	-42.5	34.7	5.7	2	II	63.2	38	68.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2131	NM_025237	Huesken	SOST	50964	ACUUGCACGAGGCCACCAGgc	CUGGUGGCCUCGUGCAAGU	-2.1	-2.2	-0.3	2	3	4	-43.7	50	2.4	5	Ib	63.2	48.9	52.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2132	NM_025237	Huesken	SOST	50964	GAGGCCACCAGGCGCACCUug	AGGUGCGCCUGGUGGCCUC	-2.1	-2.4	-4.2	0	0	5	-48.5	37.2	-1.6	1	III	73.7	40.7	71.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2133	NM_025237	Huesken	SOST	50964	CCAGGCGCACCUUGCGCGCgc	GCGCGCAAGGUGCGCCUGG	-3.4	-3.3	-8.2	-2	1	6	-47.7	26	6.1	0	II	78.9	38.4	41.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2134	NM_025237	Huesken	SOST	50964	CGCACCUUGCGCGCGCGCGgc	CGCGCGCGCGCAAGGUGCG	-2.4	-2.4	-5.9	1	0	11	-47.5	34.2	-4.4	0	II	84.2	39.2	60.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2135	NM_025237	Huesken	SOST	50964	CACCUUGCGCGCGCGCGGCgc	GCCGCGCGCGCGCAAGGUG	-3.4	-2.1	-4.6	1	1	13	-48.4	28.6	3.1	1	II	84.2	42.5	57.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2136	NM_025237	Huesken	SOST	50964	GCGCGCGCGGCGCCUCACCac	GGUGAGGCGCCGCGCGCGC	-3.3	-3.4	-5.2	2	0	14	-51.1	15.5	7.5	-1	II	89.5	34.3	26	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2137	NM_025237	Huesken	SOST	50964	GCGCGCGGCGCCUCACCACcg	GUGGUGAGGCGCCGCGCGC	-2.2	-3.4	-7.1	2	0	12	-49.6	17.7	2.7	-1	II	84.2	30	65.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2138	NM_025237	Huesken	SOST	50964	CGCGGCGCCUCACCACCGGga	CCGGUGGUGAGGCGCCGCG	-3.3	-2.4	-7.1	0	0	9	-49.4	17.4	-2	-1	II	84.2	33.8	35.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2139	NM_025237	Huesken	SOST	50964	CGCCUCACCACCGGGACACag	GUGUCCCGGUGGUGAGGCG	-2.2	-2.4	-0.6	0	-1	5	-46.7	32.5	0.7	2	II	73.7	36.9	62.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2140	NM_025237	Huesken	SOST	50964	CCACCGGGACACAGCAGCUgc	AGCUGCUGUGUCCCGGUGG	-2.1	-3.3	-3.4	-1	-1	5	-46.1	24.7	-8.6	0	III	68.4	34.5	66	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2141	NM_025237	Huesken	SOST	50964	CACGCGCUGCGCGCGGUAGcg	CUACCGCGCGCAGCGCGUG	-2.1	-2.1	-6.8	0	-1	8	-46.4	30.4	-4.4	0	II	78.9	32.3	50.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2142	NM_025237	Huesken	SOST	50964	ACGCGCUGCGCGCGGUAGCgg	GCUACCGCGCGCAGCGCGU	-3.4	-2.2	-6.8	3	2	8	-47.7	28.8	7.5	1	II	78.9	44.8	51.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2143	NM_025237	Huesken	SOST	50964	GUCGGGCCCACUAGGUCGCca	GCGACCUAGUGGGCCCGAC	-3.4	-2.2	-1.5	0	2	7	-47.1	32.5	5.5	0	II	73.7	41.4	66	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2144	NM_025237	Huesken	SOST	50964	CCACUAGGUCGCCACCACUug	AGUGGUGGCGACCUAGUGG	-2.1	-3.3	-1.8	0	-2	4	-44.1	34.2	-5.9	1	III	63.2	38.6	78.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2145	NM_025237	Huesken	SOST	50964	ACUAGGUCGCCACCACUUGcc	CAAGUGGUGGCGACCUAGU	-2.1	-2.2	-1.8	3	2	4	-41.7	49.5	0.1	4	Ib	57.9	53.3	64.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2146	NM_025237	Huesken	SOST	50964	CACCACUUGCCGCGGCCGAug	UCGGCCGCGGCAAGUGGUG	-2.4	-2.1	-6.9	1	-2	10	-46.7	28.5	-3	1	III	73.7	29.1	66	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2147	NM_025237	Huesken	SOST	50964	GCGGCCGAUGGCGUUGGGCag	GCCCAACGCCAUCGGCCGC	-3.4	-3.4	-2.1	0	2	7	-48.1	27.6	7.4	-1	II	78.9	34.5	64.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2148	NM_025237	Huesken	SOST	50964	UGGCGUUGGGCAGCAGGCGcg	CGCCUGCUGCCCAACGCCA	-2.4	-2.1	0	3	3	4	-47.3	32.8	-2.4	3	II	73.7	58.5	64.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2149	NM_025237	Huesken	SOST	50964	UUGGGCAGCAGGCGCGCCGgg	CGGCGCGCCUGCUGCCCAA	-2.4	-0.9	-3.5	0	4	9	-48.8	38.5	2.4	4	II	78.9	65.8	84.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2150	NM_025237	Huesken	SOST	50964	GCAGGCGCGCCGGGCCGCAcu	UGCGGCCCGGCGCGCCUGC	-2.1	-3.4	-3.5	0	-1	15	-52.8	-2.8	-6.1	0	III	89.5	19.6	52.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2151	NM_025237	Huesken	SOST	50964	GGGCCGCACUGGCCGGAGCac	GCUCCGGCCAGUGCGGCCC	-3.4	-3.3	-7	2	0	7	-51.2	23.8	9.8	-2	II	84.2	28.9	53	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2152	NM_025237	Huesken	SOST	50964	CACACCAGCUCGGUGACCGgc	CGGUCACCGAGCUGGUGUG	-2.4	-2.1	-3.8	0	2	3	-44.3	48.2	-2	1	II	68.4	49.6	52.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2153	NM_025237	Huesken	SOST	50964	CUCGGUGACCGGCUUGGCGcu	CGCCAAGCCGGUCACCGAG	-2.4	-2.1	-1.6	0	2	5	-45.7	41	2.7	0	II	73.7	47.1	59.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2154	NM_025237	Huesken	SOST	50964	GGUGACCGGCUUGGCGCUGcg	CAGCGCCAAGCCGGUCACC	-2.1	-3.3	-1.6	0	1	5	-46.4	21.8	5.7	0	II	73.7	28.5	50.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2155	NM_025237	Huesken	SOST	50964	GACCGGCUUGGCGCUGCGGca	CCGCAGCGCCAAGCCGGUC	-3.3	-2.4	-1.6	1	2	5	-47.9	25.5	-2.4	-1	II	78.9	35.5	51.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2156	NM_025237	Huesken	SOST	50964	CGGCUUGGCGCUGCGGCACgg	GUGCCGCAGCGCCAAGCCG	-2.2	-2.4	-3.5	1	-1	5	-47.7	29.9	6.1	0	II	78.9	36.7	49.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2157	NM_025237	Huesken	SOST	50964	GCUUGGCGCUGCGGCACGGcc	CCGUGCCGCAGCGCCAAGC	-3.3	-3.4	-3.8	0	2	5	-47.7	20.3	-2.4	0	II	78.9	28.9	50.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2158	NM_025237	Huesken	SOST	50964	GCUGCGGCACGGCCCAUCGgu	CGAUGGGCCGUGCCGCAGC	-2.4	-3.4	-2.7	0	1	6	-48.1	33.7	2.4	1	II	78.9	39.3	51	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2159	NM_025237	Huesken	SOST	50964	CUGCGGCACGGCCCAUCGGuc	CCGAUGGGCCGUGCCGCAG	-3.3	-2.1	-2.7	1	1	6	-48	25.8	-2	-1	II	78.9	41	54.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2160	NM_025237	Huesken	SOST	50964	CCCAUCGGUCACGUAGCGGgu	CCGCUACGUGACCGAUGGG	-3.3	-3.3	0.2	-1	1	4	-43.9	39.4	0.3	2	II	68.4	39.9	64.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2161	NM_025237	Huesken	SOST	50964	CCAUCGGUCACGUAGCGGGug	CCCGCUACGUGACCGAUGG	-3.3	-3.3	0.3	0	1	5	-43.9	44.6	-4.4	3	II	68.4	47	77.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2162	NM_025237	Huesken	SOST	50964	CGGUCACGUAGCGGGUGAAgu	UUCACCCGCUACGUGACCG	-0.9	-2.4	-1.9	-2	-4	5	-42.6	31.5	-8.1	0	III	63.2	15.8	57.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2163	NM_025237	Huesken	SOST	50964	GGUCACGUAGCGGGUGAAGug	CUUCACCCGCUACGUGACC	-2.1	-3.3	-1.9	0	0	5	-42.3	32.6	0.4	1	II	63.2	27.3	59.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2164	NM_025237	Huesken	SOST	50964	GUAGCGGGUGAAGUGCAGCuc	GCUGCACUUCACCCGCUAC	-3.4	-2.2	0.3	0	3	5	-43	40.9	12	2	II	63.2	50.6	75.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2165	NM_025237	Huesken	SOST	50964	GGGUGAAGUGCAGCUCGCGgc	CGCGAGCUGCACUUCACCC	-2.4	-3.3	-0.9	0	1	4	-44.3	28.9	-2.4	1	II	68.4	38.5	54.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2166	NM_025237	Huesken	SOST	50964	GGUGAAGUGCAGCUCGCGGca	CCGCGAGCUGCACUUCACC	-3.3	-3.3	-0.9	1	3	5	-44.3	38.6	9.7	2	II	68.4	38.7	66.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2167	NM_025237	Huesken	SOST	50964	GCAGCUGUACUCGGACACGuc	CGUGUCCGAGUACAGCUGC	-2.4	-3.4	-0.2	-1	1	3	-42.4	32.8	-2.4	1	II	63.2	40.7	54.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2168	NM_025237	Huesken	SOST	50964	CAGCUGUACUCGGACACGUcu	ACGUGUCCGAGUACAGCUG	-2.2	-2.1	-0.2	2	-1	3	-41.2	42.3	-8.3	2	III	57.9	45	74.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2169	NM_025237	Huesken	SOST	50964	AGCUGUACUCGGACACGUCuu	GACGUGUCCGAGUACAGCU	-2.4	-2.1	-0.2	1	2	3	-41.5	52.3	4.8	3	Ib	57.9	52.6	75.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2170	NM_025237	Huesken	SOST	50964	CUGUACUCGGACACGUCUUug	AAGACGUGUCCGAGUACAG	-0.9	-2.1	-0.2	1	-2	3	-39	58.1	-5.9	4	II	52.6	45.9	87.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2171	NM_025237	Huesken	SOST	50964	CUUUGGUCUCAAAGGGGUGgu	CACCCCUUUGAGACCAAAG	-2.1	-2.1	-2.2	-1	1	4	-38.6	52.6	-2.3	3	II	52.6	49.1	72.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2172	NM_025237	Huesken	SOST	50964	GGUGGUGGGGAGGCCGCCCuc	GGGCGGCCUCCCCACCACC	-3.3	-3.3	-3.1	1	2	8	-52	21.2	9.7	0	II	84.2	40.8	64.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2173	NM_025237	Huesken	SOST	50964	GGGAGGCCGCCCUCCGUUCuc	GAACGGAGGGCGGCCUCCC	-2.4	-3.3	-8	1	-1	8	-49.2	29.7	2.7	0	II	78.9	34.2	52.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2174	NM_025237	Huesken	SOST	50964	UCUCCGCCCGGUUCAUGGUcu	ACCAUGAACCGGGCGGAGA	-2.2	-2.4	-1.3	2	2	8	-44.6	54.7	-1	2	II	63.2	47.7	69.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2175	NM_025237	Huesken	SOST	50964	UCCGCCCGGUUCAUGGUCUug	AGACCAUGAACCGGGCGGA	-2.1	-2.4	-1.3	1	1	8	-44.6	48.2	-6.6	2	II	63.2	49.2	78.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2176	NM_025237	Huesken	SOST	50964	CCCGGUUCAUGGUCUUGUUgu	AACAAGACCAUGAACCGGG	-0.9	-3.3	-0.7	0	-3	5	-39.2	54.6	-6	1	III	52.6	31.9	69.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2177	NM_025237	Huesken	SOST	50964	CGGUUCAUGGUCUUGUUGUuc	ACAACAAGACCAUGAACCG	-2.2	-2.4	1.4	0	-2	3	-36.9	70	-6	4	III	47.4	44.1	52.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2178	NM_025237	Huesken	SOST	50964	GUUGUUCUCCAGCUCCGGUgg	ACCGGAGCUGGAGAACAAC	-2.2	-2.2	-1.8	1	2	4	-41.8	59.5	3.1	1	II	57.9	42.3	83.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2179	NM_025237	Huesken	SOST	50964	UGUUCUCCAGCUCCGGUGGag	CCACCGGAGCUGGAGAACA	-3.3	-2.1	-1.8	0	3	4	-44.1	68.2	3.1	4	Ib	63.2	58.3	86.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2180	NM_025237	Huesken	SOST	50964	UCUCCAGCUCCGGUGGAGGcu	CCUCCACCGGAGCUGGAGA	-3.3	-2.4	-5.6	1	4	4	-46.7	44.3	0	3	II	68.4	54.2	68.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2181	NM_025237	Huesken	SOST	50964	GCUCGGGGUACUCUCCGAGcu	CUCGGAGAGUACCCCGAGC	-2.1	-3.4	-6.5	1	2	5	-45.1	37	5.4	1	II	68.4	35.6	44.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2182	NM_025237	Huesken	SOST	50964	CUCUCCGAGCUCGGGGAUGau	CAUCCCCGAGCUCGGAGAG	-2.1	-2.1	-3.6	-1	-1	5	-45	35	-4.4	1	II	68.4	38.2	50.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2183	NM_025237	Huesken	SOST	50964	CCGAGCUCGGGGAUGAUUUcc	AAAUCAUCCCCGAGCUCGG	-0.9	-3.3	-2.5	1	-3	5	-41.3	50.1	-6.2	1	III	57.9	38.5	49.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2184	NM_025237	Huesken	SOST	50964	UCCGUGGCAUCAUUCUUGAac	UCAAGAAUGAUGCCACGGA	-2.4	-2.4	0.7	0	0	3	-38.7	62.4	-3.8	6	II	47.4	50.5	68.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2185	NM_025237	Huesken	SOST	50964	UGGCAUCAUUCUUGAACGCcu	GCGUUCAAGAAUGAUGCCA	-3.4	-2.1	-2.4	0	3	3	-37.3	61	10.8	3	II	47.4	58.2	93.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2186	NM_025237	Huesken	SOST	50964	CAUCAUUCUUGAACGCCUGcc	CAGGCGUUCAAGAAUGAUG	-2.1	-2.1	-0.6	1	1	4	-36	61.8	-1.9	4	II	47.4	54.3	82.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2187	NM_025237	Huesken	SOST	50964	CUUGAACGCCUGCCACCCCug	GGGGUGGCAGGCGUUCAAG	-3.3	-2.1	0	0	2	4	-44.8	36.3	10.5	2	II	68.4	48.9	48.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2188	NM_025237	Huesken	SOST	50964	AACGCCUGCCACCCCUGGCcc	GCCAGGGGUGGCAGGCGUU	-3.4	-0.9	-5.1	3	4	4	-48.2	51.1	9.8	2	II	73.7	56.4	77.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2189	NM_025237	Huesken	SOST	50964	CCUGCCACCCCUGGCCCUCca	GAGGGCCAGGGGUGGCAGG	-2.4	-3.3	-5.1	-1	0	5	-50.4	28.9	6.1	0	II	78.9	37.6	71.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2190	NM_025237	Huesken	SOST	50964	GCCACCCCUGGCCCUCCACua	GUGGAGGGCCAGGGGUGGC	-2.2	-3.4	-3.4	0	0	5	-50.5	34.8	7.5	-1	II	78.9	28.1	52.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2191	NM_025237	Huesken	SOST	50964	CACCCCUGGCCCUCCACUAca	UAGUGGAGGGCCAGGGGUG	-1.3	-2.1	-1.6	2	-3	5	-47.2	41.7	-10.5	0	III	68.4	34.1	72	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2192	NM_025237	Huesken	SOST	50964	ACCCCUGGCCCUCCACUACac	GUAGUGGAGGGCCAGGGGU	-2.2	-2.2	-1.6	3	1	5	-47.3	46.2	10.5	2	II	68.4	44.2	63.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2193	NM_025237	Huesken	SOST	50964	CCUGGCCCUCCACUACACGga	CGUGUAGUGGAGGGCCAGG	-2.4	-3.3	-1.6	-1	1	5	-45.2	35.7	0.4	-1	II	68.4	40.1	60.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2194	NM_025237	Huesken	SOST	50964	CUGGCCCUCCACUACACGGaa	CCGUGUAGUGGAGGGCCAG	-3.3	-2.1	-1.6	0	1	5	-45.2	49	-2.1	-1	II	68.4	49.5	55.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2195	NM_005450	Huesken	NOG	9241	CUGACAGCGCCACCGCAGCac	GCUGCGGUGGCGCUGUCAG	-3.4	-2.1	-2.8	0	0	5	-46.5	35.8	5.5	1	II	73.7	46.1	80.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2196	NM_005450	Huesken	NOG	9241	CCACCGCAGCACCGUGAGGug	CCUCACGGUGCUGCGGUGG	-3.3	-3.3	-0.9	0	0	4	-46.4	24.8	0.4	0	II	73.7	31.7	70.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2197	NM_005450	Huesken	NOG	9241	ACCGCAGCACCGUGAGGUGca	CACCUCACGGUGCUGCGGU	-2.1	-2.2	-3.8	0	2	4	-45.3	30.9	-2.4	1	II	68.4	37.3	56.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2198	NM_005450	Huesken	NOG	9241	GAGGUGCACGGACUUGGACgg	GUCCAAGUCCGUGCACCUC	-2.2	-2.4	1.2	1	2	3	-43.1	44.1	12.1	1	II	63.2	39	55.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2199	NM_005450	Huesken	NOG	9241	GUGCACGGACUUGGACGGCuu	GCCGUCCAAGUCCGUGCAC	-3.4	-2.2	0.9	0	2	4	-44.4	27.1	13.1	2	II	68.4	38.5	66.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2200	NM_005450	Huesken	NOG	9241	GGACUUGGACGGCUUGCACac	GUGCAAGCCGUCCAAGUCC	-2.2	-3.3	1.5	1	2	4	-42.8	47.2	12.1	2	II	63.2	37.2	75.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2201	NM_005450	Huesken	NOG	9241	UGGACGGCUUGCACACCAUgc	AUGGUGUGCAAGCCGUCCA	-1.1	-2.1	-3.6	0	0	4	-42.7	48.2	-8.9	3	II	57.9	48.5	86.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2202	NM_005450	Huesken	NOG	9241	ACGGCUUGCACACCAUGCCcu	GGCAUGGUGUGCAAGCCGU	-3.3	-2.2	-3.8	3	3	4	-43.7	52.4	7.5	2	II	63.2	59.7	79.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2203	NM_005450	Huesken	NOG	9241	GGCUUGCACACCAUGCCCUcg	AGGGCAUGGUGUGCAAGCC	-2.1	-3.3	-2.4	1	0	4	-44.5	46.2	-8.9	1	III	63.2	37.7	75.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2204	NM_005450	Huesken	NOG	9241	CGGAGCACGAGCGCUUACUga	AGUAAGCGCUCGUGCUCCG	-2.1	-2.4	-2.7	0	-3	4	-42.6	43.5	-8.3	1	III	63.2	39.6	51	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2205	NM_005450	Huesken	NOG	9241	CACGAGCGCUUACUGAAGCag	GCUUCAGUAAGCGCUCGUG	-3.4	-2.1	-2.5	0	1	4	-39.9	45.8	12.8	0	II	57.9	47.4	84.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2206	NM_005450	Huesken	NOG	9241	GCGCUUACUGAAGCAGCUGcc	CAGCUGCUUCAGUAAGCGC	-2.1	-3.4	-2.5	1	0	4	-40.5	41.4	2.3	0	II	57.9	41.7	56.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2207	NM_005450	Huesken	NOG	9241	CGCUUACUGAAGCAGCUGCcc	GCAGCUGCUUCAGUAAGCG	-3.4	-2.4	-2.5	0	-1	3	-40.5	57.1	10.1	2	II	57.9	43.3	63.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2208	NM_005450	Huesken	NOG	9241	CGCGGCCAAAAGCGGCUGCcc	GCAGCCGCUUUUGGCCGCG	-3.4	-2.4	-3.7	-2	-1	7	-45.2	31.9	10.1	0	II	73.7	34	46.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2209	NM_005450	Huesken	NOG	9241	AAAAGCGGCUGCCCAGGUCgu	GACCUGGGCAGCCGCUUUU	-2.4	-0.9	-1.7	2	4	5	-43.6	47.9	7.5	4	Ib	63.2	55.7	77.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2210	NM_005450	Huesken	NOG	9241	AAGCGGCUGCCCAGGUCGUuc	ACGACCUGGGCAGCCGCUU	-2.2	-0.9	-1.6	2	2	5	-46.4	28.2	-11.3	1	II	68.4	41.1	51.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2211	NM_005450	Huesken	NOG	9241	GCGGCUGCCCAGGUCGUUCca	GAACGACCUGGGCAGCCGC	-2.4	-3.4	-0.8	1	0	5	-46.7	38.7	12	1	II	73.7	38.8	59.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2212	NM_005450	Huesken	NOG	9241	CGGCUGCCCAGGUCGUUCCac	GGAACGACCUGGGCAGCCG	-3.3	-2.4	-0.8	1	-1	4	-46.6	46.6	5.4	2	II	73.7	45	51.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2213	NM_005450	Huesken	NOG	9241	CCAGGUCGUUCCACGCGUAca	UACGCGUGGAACGACCUGG	-1.3	-3.3	0	-1	-3	4	-42.6	26.5	-13	0	III	63.2	27.1	51.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2214	NM_005450	Huesken	NOG	9241	UCGUUCCACGCGUACAGCAcg	UGCUGUACGCGUGGAACGA	-2.1	-2.4	-0.9	1	0	4	-41.5	55	-6.1	4	II	57.9	51.8	85.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2215	NM_005450	Huesken	NOG	9241	UCCACGCGUACAGCACGGGgc	CCCGUGCUGUACGCGUGGA	-3.3	-2.4	-5.4	0	3	4	-44.8	38.3	-2.4	3	II	68.4	53.7	60.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2216	NM_005450	Huesken	NOG	9241	GCGUACAGCACGGGGCAGAau	UCUGCCCCGUGCUGUACGC	-2.4	-3.4	-0.1	0	-3	6	-45.7	26	-3.4	3	III	68.4	24.9	61	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2217	NM_005450	Huesken	NOG	9241	GUACAGCACGGGGCAGAAUgu	AUUCUGCCCCGUGCUGUAC	-1.1	-2.2	0.2	2	-1	6	-41.9	33.2	-1.3	2	III	57.9	30.7	51	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2218	NM_005450	Huesken	NOG	9241	ACAGCACGGGGCAGAAUGUcu	ACAUUCUGCCCCGUGCUGU	-2.2	-2.2	0.2	1	1	6	-42.7	33.9	-8.9	3	II	57.9	38.9	60	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2219	NM_005450	Huesken	NOG	9241	ACGGGGCAGAAUGUCUGCGac	CGCAGACAUUCUGCCCCGU	-2.4	-2.2	-5.2	2	4	6	-43.2	43.5	7.7	1	II	63.2	52.6	53	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2220	NM_005450	Huesken	NOG	9241	CGGGGCAGAAUGUCUGCGAcc	UCGCAGACAUUCUGCCCCG	-2.4	-2.4	-5.2	-1	-3	6	-43.4	37.1	-5.8	0	III	63.2	34	47.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2221	NM_005450	Huesken	NOG	9241	GGCAGAAUGUCUGCGACCAca	UGGUCGCAGACAUUCUGCC	-2.1	-3.3	-5.2	1	-2	3	-42	40.3	-3.1	0	III	57.9	29.8	40.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2222	NM_005450	Huesken	NOG	9241	CAGAAUGUCUGCGACCACAgc	UGUGGUCGCAGACAUUCUG	-2.1	-2.1	-0.9	0	-3	3	-39.6	50.3	-5.4	4	II	52.6	45	57.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2223	NM_005450	Huesken	NOG	9241	UGUCUGCGACCACAGCCACau	GUGGCUGUGGUCGCAGACA	-2.2	-2.1	-5.8	1	3	3	-44.1	45	7.5	3	II	63.2	53.1	83.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2224	NM_005450	Huesken	NOG	9241	CUGCGACCACAGCCACAUCug	GAUGUGGCUGUGGUCGCAG	-2.4	-2.1	-6.8	0	-1	3	-43	38.7	10.1	-1	II	63.2	40.2	81.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2225	NM_005450	Huesken	NOG	9241	UAACUUCCUCCGCAGCUUCuu	GAAGCUGCGGAGGAAGUUA	-2.4	-1.3	-1.5	1	3	4	-39.7	66.5	7.5	6	Ia	52.6	68.8	102	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2226	NM_005450	Huesken	NOG	9241	UCCUCCGCAGCUUCUUGCUua	AGCAAGAAGCUGCGGAGGA	-2.1	-2.4	-2.5	0	1	4	-42.9	66.5	-3.3	3	II	57.9	51.6	97.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2227	NM_005450	Huesken	NOG	9241	CGCAGCUUCUUGCUUAGGCgc	GCCUAAGCAAGAAGCUGCG	-3.4	-2.4	-2.5	1	1	3	-40.4	59.5	10.1	0	II	57.9	46.6	85.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2228	NM_005450	Huesken	NOG	9241	GCUUCUUGCUUAGGCGCUGcu	CAGCGCCUAAGCAAGAAGC	-2.1	-3.4	-2.1	1	1	5	-40.4	46.1	0	1	II	57.9	39	62.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2229	NM_005450	Huesken	NOG	9241	CUUGCUUAGGCGCUGCUUCuu	GAAGCAGCGCCUAAGCAAG	-2.4	-2.1	-2.1	1	1	5	-40.4	63.4	7.8	2	II	57.9	55.1	83	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2230	NM_005450	Huesken	NOG	9241	UUAGGCGCUGCUUCUUGCCcu	GGCAAGAAGCAGCGCCUAA	-3.3	-0.9	-2.9	0	5	5	-41.6	65.5	8.1	4	II	57.9	73.2	91.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2231	NM_005450	Huesken	NOG	9241	UAGGCGCUGCUUCUUGCCCug	GGGCAAGAAGCAGCGCCUA	-3.3	-1.3	-4	1	5	5	-44	61.1	15.2	2	II	63.2	67.6	93.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2232	NM_005450	Huesken	NOG	9241	GCUUCUUGCCCUGGGCCAAgc	UUGGCCCAGGGCAAGAAGC	-0.9	-3.4	-2.7	1	-2	5	-44.4	42.1	-3.1	1	II	63.2	22.2	41.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2233	NM_005450	Huesken	NOG	9241	CAAGCCCUCGGAGAACUCUag	AGAGUUCUCCGAGGGCUUG	-2.1	-2.1	-4.5	0	-1	4	-41.9	44.8	-6	1	III	57.9	44	55.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2234	NM_005450	Huesken	NOG	9241	GAACUCUAGCCCUUUGAUCuc	GAUCAAAGGGCUAGAGUUC	-2.4	-2.4	-0.8	3	2	4	-37.4	68.8	5	4	II	47.4	52.2	78.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2235	NM_005450	Huesken	NOG	9241	UCUAGCCCUUUGAUCUCGCuc	GCGAGAUCAAAGGGCUAGA	-3.4	-2.4	0.4	0	5	4	-40.1	63.4	7.1	5	II	52.6	63.5	96.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2236	NM_005450	Huesken	NOG	9241	CCCCGACGGCCGCUGCCGCag	GCGGCAGCGGCCGUCGGGG	-3.4	-3.3	-7	1	1	7	-51.8	12.6	7.8	-2	II	89.5	28.4	70.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2237	NM_005450	Huesken	NOG	9241	CCGCUGCCGCAGCAGCUGGuc	CCAGCUGCUGCGGCAGCGG	-3.3	-3.3	-9.1	1	-1	5	-48.5	26.5	2.7	0	II	78.9	35.5	78.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2238	NM_005450	Huesken	NOG	9241	GCUGCCGCAGCAGCUGGUCca	GACCAGCUGCUGCGGCAGC	-2.4	-3.4	-7.4	0	1	5	-47.4	28.9	5.1	0	II	73.7	33.3	55.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2239	NM_005450	Huesken	NOG	9241	GCAGCAGCUGGUCCAGCUCcg	GAGCUGGACCAGCUGCUGC	-2.4	-3.4	-4.6	0	1	2	-46.1	36.1	12.8	0	II	68.4	37.6	71.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2240	NM_005450	Huesken	NOG	9241	GCUGGUCCAGCUCCGCCAGgu	CUGGCGGAGCUGGACCAGC	-2.1	-3.4	-2.7	0	1	5	-47.5	33.9	3.1	-1	II	73.7	30.7	58.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2241	NM_005450	Huesken	NOG	9241	GCGAGGUGGCCAUGAAGCCug	GGCUUCAUGGCCACCUCGC	-3.3	-3.4	-5.7	1	1	4	-45.3	29.3	5.1	0	II	68.4	39.6	64.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2242	NM_005450	Huesken	NOG	9241	UGGCCAUGAAGCCUGGGUCgu	GACCCAGGCUUCAUGGCCA	-2.4	-2.1	-5.9	2	3	4	-44.9	58.8	9.8	3	II	63.2	59.1	82.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2243	NM_005450	Huesken	NOG	9241	CCCGAGCAGCGAGCGCAGCag	GCUGCGCUCGCUGCUCGGG	-3.4	-3.3	-1.3	0	-1	4	-48.2	15.6	8.1	-1	II	78.9	32.4	67.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2244	NM_005450	Huesken	NOG	9241	GAGCAGCGAGCGCAGCAGCgu	GCUGCUGCGCUCGCUGCUC	-3.4	-2.4	-1.5	1	1	4	-46.8	26.1	10.2	0	II	73.7	39.4	78.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2245	NM_005450	Huesken	NOG	9241	AGCAGCGAGCGCAGCAGCGuc	CGCUGCUGCGCUCGCUGCU	-2.4	-2.1	-2.6	1	3	4	-46.8	29.9	-4.9	2	II	73.7	47.8	73.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2246	NM_005450	Huesken	NOG	9241	CAGCGAGCGCAGCAGCGUCuc	GACGCUGCUGCGCUCGCUG	-2.4	-2.1	-3.1	1	0	4	-45.9	27.3	10.1	0	II	73.7	40.9	70.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2247	NM_005450	Huesken	NOG	9241	GCGCAGCAGCGUCUCGUUCag	GAACGAGACGCUGCUGCGC	-2.4	-3.4	-3.1	2	0	4	-43.8	37.9	17.9	0	II	68.4	33.5	74.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2248	NM_005450	Huesken	NOG	9241	GCAGCAGCGUCUCGUUCAGau	CUGAACGAGACGCUGCUGC	-2.1	-3.4	-0.1	0	1	3	-42.2	34.1	3.1	1	II	63.2	26.9	46.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2249	NM_005450	Huesken	NOG	9241	GUUCAGAUCCUUUUCCUUGgg	CAAGGAAAAGGAUCUGAAC	-2.1	-2.2	0.1	2	2	2	-34.5	75.2	8.1	5	II	42.1	56.9	95.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2250	NM_005450	Huesken	NOG	9241	CCUUGGGGUCAAAGAUAGGgu	CCUAUCUUUGACCCCAAGG	-3.3	-3.3	-1.7	-2	0	4	-39.1	40.5	-2.3	2	II	52.6	39.9	89.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2251	NM_005450	Huesken	NOG	9241	GGGUCAAAGAUAGGGUCUGgg	CAGACCCUAUCUUUGACCC	-2.1	-3.3	0.6	0	0	3	-39.5	41.8	0	3	II	52.6	37.8	88.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2252	NM_005450	Huesken	NOG	9241	GGGUCUGGGUGUUCGAUGAgg	UCAUCGAACACCCAGACCC	-2.4	-3.3	1.5	1	-3	3	-42.2	55.8	-0.7	2	III	57.9	28.9	67.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2253	NM_005450	Huesken	NOG	9241	GUCUGGGUGUUCGAUGAGGuc	CCUCAUCGAACACCCAGAC	-3.3	-2.2	1.5	1	1	3	-41	42.6	-2.4	2	II	57.9	43.6	79.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2254	NM_005450	Huesken	NOG	9241	UCUGGGUGUUCGAUGAGGUcc	ACCUCAUCGAACACCCAGA	-2.2	-2.4	2	1	3	3	-41	54	-6.6	5	II	52.6	56	101.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2255	NM_005450	Huesken	NOG	9241	UGUUCGAUGAGGUCCACCAgg	UGGUGGACCUCAUCGAACA	-2.1	-2.1	-2.3	1	1	2	-41	71	-3.8	6	II	52.6	61	96.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2256	NM_005450	Huesken	NOG	9241	UUCGAUGAGGUCCACCAGGgg	CCUGGUGGACCUCAUCGAA	-3.3	-0.9	-2.4	1	3	2	-42.1	51.6	2.8	4	Ib	57.9	70.5	81.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2257	NM_005450	Huesken	NOG	9241	GGGGCAGGUUGUCGCUGGGug	CCCAGCGACAACCUGCCCC	-3.3	-3.3	-0.9	-1	1	5	-47.2	28	5.4	0	II	73.7	27.9	68.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2258	NM_005450	Huesken	NOG	9241	UUGUCGCUGGGUGCCGGGCgg	GCCCGGCACCCAGCGACAA	-3.4	-0.9	-1	1	4	7	-47.6	50.7	10.4	3	II	73.7	61	88.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2259	NM_005450	Huesken	NOG	9241	UCGCUGGGUGCCGGGCGGAug	UCCGCCCGGCACCCAGCGA	-2.4	-2.4	-1.7	0	0	9	-50.5	15.5	-8.5	3	II	78.9	39.1	72	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2260	NM_005450	Huesken	NOG	9241	CGCUGGGUGCCGGGCGGAUgu	AUCCGCCCGGCACCCAGCG	-1.1	-2.4	-1	-1	-3	9	-49.2	12.9	-8.3	0	III	78.9	17.3	42.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2261	NM_005450	Huesken	NOG	9241	GGGUGCCGGGCGGAUGUGGag	CCACAUCCGCCCGGCACCC	-3.3	-3.3	-0.2	1	0	9	-48.9	22.5	-2.4	1	II	78.9	29.2	51.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2262	NM_005450	Huesken	NOG	9241	GCCGGGCGGAUGUGGAGAUag	AUCUCCACAUCCGCCCGGC	-1.1	-3.4	-0.2	0	-2	9	-46	20.3	-4	0	III	68.4	18.7	60.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2263	NM_005450	Huesken	NOG	9241	GGGCGGAUGUGGAGAUAGUgc	ACUAUCUCCACAUCCGCCC	-2.2	-3.3	4.4	1	-2	6	-42.5	29.8	-6.6	0	III	57.9	26.3	52.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2264	NM_005450	Huesken	NOG	9241	GGAGAUAGUGCUGGCCGCCgg	GGCGGCCAGCACUAUCUCC	-3.3	-3.3	-1.7	-1	2	7	-45.7	30.8	10.8	1	II	68.4	44.6	87.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2265	NM_005450	Huesken	NOG	9241	GAUAGUGCUGGCCGCCGGCcg	GCCGGCGGCCAGCACUAUC	-3.4	-2.4	-1.8	-1	3	10	-47	37.7	7.5	2	II	73.7	45.2	82.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2266	NM_007019	Huesken	UBE2C	11065	UUCCAGAGCUCGGCAGCAUgu	AUGCUGCCGAGCUCUGGAA	-1.1	-0.9	-1.8	2	1	4	-43.1	42.6	-3.6	4	II	57.9	50.1	84.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2267	NM_007019	Huesken	UBE2C	11065	UCCAGAGCUCGGCAGCAUGug	CAUGCUGCCGAGCUCUGGA	-2.1	-2.4	-1.8	0	1	4	-44.3	45.1	2.4	3	II	63.2	53.7	90.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2268	NM_007019	Huesken	UBE2C	11065	AGCUCGGCAGCAUGUGUGUuc	ACACACAUGCUGCCGAGCU	-2.2	-2.1	-1.8	0	0	4	-42.8	48.8	-6.3	3	II	57.9	39.8	78	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2269	NM_007019	Huesken	UBE2C	11065	CUCGGCAGCAUGUGUGUUCaa	GAACACACAUGCUGCCGAG	-2.4	-2.1	2.4	-1	-1	4	-40.6	51.5	5.4	3	II	57.9	49.3	89.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2270	NM_007019	Huesken	UBE2C	11065	UCGGCAGCAUGUGUGUUCAag	UGAACACACAUGCUGCCGA	-2.1	-2.4	2.1	0	0	4	-40.6	58.7	1.6	5	II	52.6	47.5	86.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2271	NM_007019	Huesken	UBE2C	11065	AUGUGUGUUCAAGGGACUAuc	UAGUCCCUUGAACACACAU	-1.3	-1.1	1.6	0	0	3	-37	56.5	-1.5	5	II	42.1	48.4	76.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2272	NM_007019	Huesken	UBE2C	11065	UUCAAGGGACUAUCAAUGUug	ACAUUGAUAGUCCCUUGAA	-2.2	-0.9	1.9	0	1	3	-34.9	70.3	-1.3	8	II	36.8	65	90.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2273	NM_007019	Huesken	UBE2C	11065	UCAAGGGACUAUCAAUGUUgg	AACAUUGAUAGUCCCUUGA	-0.9	-2.4	1.9	1	1	3	-34.9	65.2	3.8	6	II	36.8	56.1	93.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2274	NM_007019	Huesken	UBE2C	11065	CUAUCAAUGUUGGGUUCUCcu	GAGAACCCAACAUUGAUAG	-2.4	-2.1	1.8	-1	1	3	-34.9	59.9	5.4	4	II	42.1	49.2	66.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2275	NM_007019	Huesken	UBE2C	11065	AUCAAUGUUGGGUUCUCCUag	AGGAGAACCCAACAUUGAU	-2.1	-1.1	-1.1	1	3	3	-36.9	78.7	1.1	5	II	42.1	63.1	94.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2276	NM_007019	Huesken	UBE2C	11065	AAUGUUGGGUUCUCCUAGAag	UCUAGGAGAACCCAACAUU	-2.4	-0.9	-1.1	2	1	3	-37.1	66.7	-6.1	5	II	42.1	54	88.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2277	NM_007019	Huesken	UBE2C	11065	UGUUGGGUUCUCCUAGAAGgc	CUUCUAGGAGAACCCAACA	-2.1	-2.1	-1.1	0	3	3	-38.1	66.3	2.4	5	II	47.4	59.3	86.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2278	NM_007019	Huesken	UBE2C	11065	UUGGGUUCUCCUAGAAGGCuc	GCCUUCUAGGAGAACCCAA	-3.4	-0.9	-1.1	0	4	3	-40.5	60.6	5.5	4	Ib	52.6	70.4	92.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2279	NM_007019	Huesken	UBE2C	11065	GUUCUCCUAGAAGGCUCUGga	CAGAGCCUUCUAGGAGAAC	-2.1	-2.2	-0.9	0	2	3	-39.6	51	2.3	3	II	52.6	41.2	75.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2280	NM_007019	Huesken	UBE2C	11065	UCCUAGAAGGCUCUGGAUGga	CAUCCAGAGCCUUCUAGGA	-2.1	-2.4	-0.7	1	2	3	-40.9	62.7	8.1	6	Ib	52.6	59.5	78.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2281	NM_007019	Huesken	UBE2C	11065	UAGAAGGCUCUGGAUGGAGag	CUCCAUCCAGAGCCUUCUA	-2.1	-1.3	-0.5	0	3	3	-40.9	51.4	2.7	7	Ib	52.6	62.2	81.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2282	NM_007019	Huesken	UBE2C	11065	CUGGAUGGAGAGCAGAAUGgu	CAUUCUGCUCUCCAUCCAG	-2.1	-2.1	1.4	-1	-1	2	-39.5	50.7	5.4	3	II	52.6	51.5	79.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2283	NM_007019	Huesken	UBE2C	11065	GGAGAGCAGAAUGGUCCUGac	CAGGACCAUUCUGCUCUCC	-2.1	-3.3	-1.2	0	1	2	-41.8	35	8	1	II	57.9	36.2	45.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2284	NM_007019	Huesken	UBE2C	11065	AGCAGAAUGGUCCUGACAUca	AUGUCAGGACCAUUCUGCU	-1.1	-2.1	-1.3	2	1	2	-39.4	56.4	-1.6	3	II	47.4	45.8	56.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2285	NM_007019	Huesken	UBE2C	11065	GCAGAAUGGUCCUGACAUCau	GAUGUCAGGACCAUUCUGC	-2.4	-3.4	-1.3	1	0	2	-39.7	39.1	5.4	2	II	52.6	37.3	48.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2286	NM_007019	Huesken	UBE2C	11065	GAAUGGUCCUGACAUCAUAca	UAUGAUGUCAGGACCAUUC	-1.3	-2.4	0.6	2	-2	2	-36.6	58.2	-1.4	4	II	42.1	39.6	32.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2287	NM_007019	Huesken	UBE2C	11065	GGUCCUGACAUCAUACAGGgc	CCUGUAUGAUGUCAGGACC	-3.3	-3.3	-4.5	2	2	2	-39.8	55.6	-2.6	2	II	52.6	49.4	42	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2288	NM_007019	Huesken	UBE2C	11065	CCUGACAUCAUACAGGGCAga	UGCCCUGUAUGAUGUCAGG	-2.1	-3.3	-3.7	0	-2	4	-40.7	47.1	-0.7	3	III	52.6	42.6	50.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2289	NM_007019	Huesken	UBE2C	11065	ACAUCAUACAGGGCAGACCac	GGUCUGCCCUGUAUGAUGU	-3.3	-2.2	-0.7	3	3	4	-40.8	61.1	7.4	6	Ia	52.6	60.8	55.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2290	NM_007019	Huesken	UBE2C	11065	CAUCAUACAGGGCAGACCAcu	UGGUCUGCCCUGUAUGAUG	-2.1	-2.1	-0.7	0	-1	4	-40.7	53.2	-0.7	4	II	52.6	46.6	28.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2291	NM_007019	Huesken	UBE2C	11065	UCAUACAGGGCAGACCACUuu	AGUGGUCUGCCCUGUAUGA	-2.1	-2.4	-0.7	1	1	4	-41.8	47.9	-3.6	7	II	52.6	60.4	61.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2292	NM_007019	Huesken	UBE2C	11065	GGCAGACCACUUUUCCUUCag	GAAGGAAAAGUGGUCUGCC	-2.4	-3.3	0.1	0	0	3	-39	59.2	12.8	2	II	52.6	39.2	63.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2293	NM_007019	Huesken	UBE2C	11065	GACCACUUUUCCUUCAGGAug	UCCUGAAGGAAAAGUGGUC	-2.4	-2.4	-2.1	2	0	2	-38	62.9	-3.4	2	III	47.4	38.6	71.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2294	NM_007019	Huesken	UBE2C	11065	CACUUUUCCUUCAGGAUGUcc	ACAUCCUGAAGGAAAAGUG	-2.2	-2.1	-2.4	0	-2	2	-35.5	70	-10.9	4	II	42.1	48.5	46.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2295	NM_007019	Huesken	UBE2C	11065	UUCCUUCAGGAUGUCCAGGca	CCUGGACAUCCUGAAGGAA	-3.3	-0.9	-2.4	2	4	2	-40.5	68.4	7.7	4	Ib	52.6	67.1	74.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2296	NM_007019	Huesken	UBE2C	11065	UCCUUCAGGAUGUCCAGGCau	GCCUGGACAUCCUGAAGGA	-3.4	-2.4	-1.6	1	3	3	-43	65.4	7.4	5	Ib	57.9	66.7	85.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2297	NM_007019	Huesken	UBE2C	11065	CCUUCAGGAUGUCCAGGCAua	UGCCUGGACAUCCUGAAGG	-2.1	-3.3	-1	-1	-2	3	-42.7	46.7	-0.3	1	III	57.9	27.3	49.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2298	NM_007019	Huesken	UBE2C	11065	AGGAUGUCCAGGCAUAUGUua	ACAUAUGCCUGGACAUCCU	-2.2	-2.1	-1	3	0	3	-39.7	69.7	-1.6	6	II	47.4	56.9	75.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2299	NM_007019	Huesken	UBE2C	11065	GGCAUAUGUUACCCUGGGUgu	ACCCAGGGUAACAUAUGCC	-2.2	-3.3	-1.5	1	-1	3	-40.6	54.7	-1.6	2	II	52.6	32.2	39.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2300	NM_007019	Huesken	UBE2C	11065	UUACCCUGGGUGUCCACGUug	ACGUGGACACCCAGGGUAA	-2.2	-0.9	-1.8	3	3	3	-42.9	60.5	-4	4	II	57.9	66.4	83	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2301	NM_007019	Huesken	UBE2C	11065	GGGUGUCCACGUUGGGGUGau	CACCCCAACGUGGACACCC	-2.1	-3.3	-1.1	-1	0	4	-44.9	41.8	3	-1	II	68.4	28.3	11.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2302	NM_007019	Huesken	UBE2C	11065	GUCCACGUUGGGGUGAUAGca	CUAUCACCCCAACGUGGAC	-2.1	-2.2	-1.9	1	1	4	-40.9	51.6	5	2	II	57.9	40.6	77.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2303	NM_007019	Huesken	UBE2C	11065	UCCACGUUGGGGUGAUAGCag	GCUAUCACCCCAACGUGGA	-3.4	-2.4	-1.9	1	2	4	-42.1	60.6	7.4	3	II	57.9	58.2	89.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2304	NM_007019	Huesken	UBE2C	11065	CGUUGGGGUGAUAGCAGGGcg	CCCUGCUAUCACCCCAACG	-3.3	-2.4	2.9	-2	1	4	-42.9	39.5	0.7	2	II	63.2	42.2	53.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2305	NM_007019	Huesken	UBE2C	11065	GUUGGGGUGAUAGCAGGGCgu	GCCCUGCUAUCACCCCAAC	-3.4	-2.2	1.1	1	3	4	-43.9	35.2	7.4	3	II	63.2	49.3	63.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2306	NM_007019	Huesken	UBE2C	11065	UAGCAGGGCGUGAGGAACUuc	AGUUCCUCACGCCCUGCUA	-2.1	-1.3	0.8	1	1	5	-43.1	41	-3.9	5	II	57.9	60.7	84.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2307	NM_007019	Huesken	UBE2C	11065	AGCAGGGCGUGAGGAACUUca	AAGUUCCUCACGCCCUGCU	-0.9	-2.1	0.8	1	0	5	-42.7	29.9	-4	3	II	57.9	36	51	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2308	NM_007019	Huesken	UBE2C	11065	GCAGGGCGUGAGGAACUUCac	GAAGUUCCUCACGCCCUGC	-2.4	-3.4	0.8	0	0	5	-43	26.8	9.7	1	II	63.2	36.9	40.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2309	NM_007019	Huesken	UBE2C	11065	GUGAGGAACUUCACUGUGGgc	CCACAGUGAAGUUCCUCAC	-3.3	-2.2	-2	1	1	2	-39.1	51.9	-4.9	3	II	52.6	52.3	74.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2310	NM_007019	Huesken	UBE2C	11065	GAGGAACUUCACUGUGGGCgc	GCCCACAGUGAAGUUCCUC	-3.4	-2.4	1.8	-1	2	4	-41.5	45.5	10.1	1	II	57.9	39.2	46.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2311	NM_007019	Huesken	UBE2C	11065	GGAACUUCACUGUGGGCGCau	GCGCCCACAGUGAAGUUCC	-3.4	-3.3	-0.6	0	2	6	-42.8	50.7	7.4	2	II	63.2	45.1	70.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2312	NM_007019	Huesken	UBE2C	11065	ACUUCACUGUGGGCGCAUUgu	AAUGCGCCCACAGUGAAGU	-0.9	-2.2	-0.6	2	0	6	-40.3	52.5	-4	4	II	52.6	35.9	57.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2313	NM_007019	Huesken	UBE2C	11065	UUCACUGUGGGCGCAUUGUaa	ACAAUGCGCCCACAGUGAA	-2.2	-0.9	-0.6	1	1	6	-40.3	66.6	-6.3	7	II	52.6	60.2	93.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2314	NM_007019	Huesken	UBE2C	11065	GUGGGCGCAUUGUAAGGGUag	ACCCUUACAAUGCGCCCAC	-2.2	-2.2	2.5	0	0	6	-41.6	38.3	-4	1	III	57.9	36.4	47.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2315	NM_007019	Huesken	UBE2C	11065	CGCAUUGUAAGGGUAGCCAcu	UGGCUACCCUUACAAUGCG	-2.1	-2.4	-0.4	-1	-2	3	-39.5	56.1	-3.4	4	III	52.6	39.3	61.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2316	NM_007019	Huesken	UBE2C	11065	GCAUUGUAAGGGUAGCCACug	GUGGCUACCCUUACAAUGC	-2.2	-3.4	-0.2	1	1	3	-39.3	56.9	7.4	4	II	52.6	44.4	87.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2317	NM_007019	Huesken	UBE2C	11065	CAUUGUAAGGGUAGCCACUgg	AGUGGCUACCCUUACAAUG	-2.1	-2.1	-0.4	-1	-1	3	-38	59	-5.5	4	II	47.4	48.9	76.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2318	NM_007019	Huesken	UBE2C	11065	UGUAAGGGUAGCCACUGGGga	CCCAGUGGCUACCCUUACA	-3.3	-2.1	-0.4	0	4	3	-42.6	57.1	2.8	7	Ib	57.9	65.4	85.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2319	NM_007019	Huesken	UBE2C	11065	GUAAGGGUAGCCACUGGGGaa	CCCCAGUGGCUACCCUUAC	-3.3	-2.2	0.4	0	3	4	-43.8	41	-7.3	3	II	63.2	49.3	67.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2320	NM_007019	Huesken	UBE2C	11065	GUAGCCACUGGGGAACUCUag	AGAGUUCCCCAGUGGCUAC	-2.1	-2.2	0.2	0	0	4	-42.8	45.3	-4	3	III	57.9	45.8	87	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2321	NM_007019	Huesken	UBE2C	11065	UAGCCACUGGGGAACUCUAgc	UAGAGUUCCCCAGUGGCUA	-1.3	-1.3	0.2	2	0	4	-41.9	51.2	-6.3	4	II	52.6	50	67.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2322	NM_007019	Huesken	UBE2C	11065	GCCACUGGGGAACUCUAGCga	GCUAGAGUUCCCCAGUGGC	-3.4	-3.4	-0.1	1	1	4	-44	49.4	14.4	1	II	63.2	41.8	69	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2323	NM_007019	Huesken	UBE2C	11065	CCACUGGGGAACUCUAGCGag	CGCUAGAGUUCCCCAGUGG	-2.4	-3.3	-1.1	1	1	4	-43	45.6	-2.1	2	II	63.2	50.1	86.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2324	NM_007019	Huesken	UBE2C	11065	GAACUCUAGCGAGAGCUUAua	UAAGCUCUCGCUAGAGUUC	-1.3	-2.4	-4	3	-2	3	-38	50.2	-3.8	4	II	47.4	38.2	60.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2325	NM_007019	Huesken	UBE2C	11065	UCUAGCGAGAGCUUAUACCuc	GGUAUAAGCUCUCGCUAGA	-3.3	-2.4	-1.8	1	4	3	-38.3	75.1	7.4	6	II	47.4	70	91.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2326	NM_007019	Huesken	UBE2C	11065	CUAGCGAGAGCUUAUACCUca	AGGUAUAAGCUCUCGCUAG	-2.1	-2.1	-1.8	-1	0	3	-38	59.6	-5.3	2	III	47.4	53.8	92.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2327	NM_007019	Huesken	UBE2C	11065	AGCGAGAGCUUAUACCUCAgg	UGAGGUAUAAGCUCUCGCU	-2.1	-2.1	-2.1	2	-1	3	-39.1	49.6	-1.1	4	II	47.4	50	80.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2328	NM_007019	Huesken	UBE2C	11065	GCGAGAGCUUAUACCUCAGgu	CUGAGGUAUAAGCUCUCGC	-2.1	-3.4	-2.1	0	0	3	-39.1	48.1	2.7	1	II	52.6	36.2	75.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2329	NM_007019	Huesken	UBE2C	11065	GAGAGCUUAUACCUCAGGUcu	ACCUGAGGUAUAAGCUCUC	-2.2	-2.4	-1.2	1	1	2	-38.8	67	0.7	2	III	47.4	44.3	80.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2330	NM_007019	Huesken	UBE2C	11065	GAGCUUAUACCUCAGGUCUuc	AGACCUGAGGUAUAAGCUC	-2.1	-2.4	-0.4	1	-1	2	-38.8	66.4	-0.9	4	II	47.4	47.5	71.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2331	NM_007019	Huesken	UBE2C	11065	AGCUUAUACCUCAGGUCUUca	AAGACCUGAGGUAUAAGCU	-0.9	-2.1	-0.4	2	0	2	-37.3	60.3	-8.9	5	II	42.1	46.1	46.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2332	NM_007019	Huesken	UBE2C	11065	CCUCAGGUCUUCAUAUACUgu	AGUAUAUGAAGACCUGAGG	-2.1	-3.3	1.8	1	-1	2	-36.5	55.3	-5.9	3	III	42.1	43.8	77.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2333	NM_007019	Huesken	UBE2C	11065	CUCAGGUCUUCAUAUACUGuu	CAGUAUAUGAAGACCUGAG	-2.1	-2.1	0.3	0	0	2	-35.3	64.8	-4.4	3	II	42.1	54.6	82.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2334	NM_007019	Huesken	UBE2C	11065	UCAGGUCUUCAUAUACUGUuc	ACAGUAUAUGAAGACCUGA	-2.2	-2.4	-0.2	0	2	2	-35.4	74.3	1.1	5	II	36.8	60.4	88.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2335	NM_007019	Huesken	UBE2C	11065	AGGUCUUCAUAUACUGUUCca	GAACAGUAUAUGAAGACCU	-2.4	-2.1	0.8	2	1	2	-34.2	87.9	10.1	5	Ib	36.8	59.9	86.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2336	NM_007019	Huesken	UBE2C	11065	CUUCAUAUACUGUUCCAGCug	GCUGGAACAGUAUAUGAAG	-3.4	-2.1	-1.2	0	2	2	-35.1	75.2	10.4	3	II	42.1	61.9	90.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2337	NM_007019	Huesken	UBE2C	11065	UUCAUAUACUGUUCCAGCUgc	AGCUGGAACAGUAUAUGAA	-2.1	-0.9	-1.5	2	2	2	-35.1	89.4	-1	6	II	36.8	69.4	97.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2338	NM_007019	Huesken	UBE2C	11065	UAUACUGUUCCAGCUGCUCca	GAGCAGCUGGAACAGUAUA	-2.4	-1.3	-1.5	0	4	2	-38.6	73.1	5.1	6	Ia	47.4	70.8	85.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2339	NM_007019	Huesken	UBE2C	11065	ACUGUUCCAGCUGCUCCAUgg	AUGGAGCAGCUGGAACAGU	-1.1	-2.2	-0.8	1	1	2	-41.4	53.8	-3.3	4	II	52.6	40.7	84.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2340	NM_007019	Huesken	UBE2C	11065	CCAGCUGCUCCAUGGAUGGuc	CCAUCCAUGGAGCAGCUGG	-3.3	-3.3	-1.2	-2	0	2	-43.8	41.8	-4.4	1	II	63.2	43.5	80.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2341	NM_007019	Huesken	UBE2C	11065	CAGCUGCUCCAUGGAUGGUcc	ACCAUCCAUGGAGCAGCUG	-2.2	-2.1	-1.7	0	-1	2	-42.7	42.3	-0.7	0	III	57.9	33.1	62.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2342	NM_002559	Huesken	P2RX3	5024	UGGGUUGGUCAAAGCCGCGau	CGCGGCUUUGACCAACCCA	-2.4	-2.1	-1	-1	3	6	-42.7	43.6	-0.3	3	II	63.2	62.5	84.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2343	NM_002559	Huesken	P2RX3	5024	CACUCCCACAGAAGUAAAGgc	CUUUACUUCUGUGGGAGUG	-2.1	-2.1	-4	-1	-1	3	-36.7	60	-4.6	2	II	47.4	41	67.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2344	NM_002559	Huesken	P2RX3	5024	AAGCCUUCAGGAGGGUGCGgu	CGCACCCUCCUGAAGGCUU	-2.4	-0.9	-0.7	2	4	3	-43.8	48.6	0	3	Ib	63.2	56.4	67.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2345	NM_002559	Huesken	P2RX3	5024	AACCUGAAGUUGUAGCCUGgg	CAGGCUACAACUUCAGGUU	-2.1	-0.9	0.9	3	3	3	-37.7	57.9	0	5	Ib	47.4	60.2	95.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2346	NM_002559	Huesken	P2RX3	5024	AAACGCUGUCGAGCCGGGUga	ACCCGGCUCGACAGCGUUU	-2.2	-0.9	-1.7	3	3	6	-43.4	40.2	-4	3	II	63.2	50.2	94.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2347	NM_002559	Huesken	P2RX3	5024	UCGAGCCGGGUGAAGGAGUau	ACUCCUUCACCCGGCUCGA	-2.2	-2.4	0.7	1	0	6	-44.7	36.7	-6.6	3	II	63.2	49.4	80.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2348	NM_002559	Huesken	P2RX3	5024	CCCAGGCCUUGUCCAAGUCgc	GACUUGGACAAGGCCUGGG	-2.4	-3.3	-0.5	-1	-1	4	-43.5	42.1	10.8	0	II	63.2	35	93	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2349	NM_002559	Huesken	P2RX3	5024	UUGACCACGUCCCCUACCCgc	GGGUAGGGGACGUGGUCAA	-3.3	-0.9	-0.6	1	4	4	-44.3	63.5	5.1	5	Ib	63.2	72	94.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2350	NM_002559	Huesken	P2RX3	5024	AGAUGGGGCAGAAAGGGUCcu	GACCCUUUCUGCCCCAUCU	-2.4	-2.1	1.8	2	3	5	-42.7	45.1	4.8	5	II	57.9	57.6	65.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2351	NM_002559	Huesken	P2RX3	5024	UGUCCGGGUGGAAGCGGCAgg	UGCCGCUUCCACCCGGACA	-2.1	-2.1	-2.1	0	1	5	-46.6	31.3	-6.3	3	II	68.4	46.4	93.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2352	NM_002559	Huesken	P2RX3	5024	AACGGAUGCUGUUCUUGAUga	AUCAAGAACAGCAUCCGUU	-1.1	-0.9	1.2	3	1	3	-36.1	69.1	-1	4	II	42.1	44.5	70.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2353	NM_002559	Huesken	P2RX3	5024	UUGAUGAAAAUAGUGAAGUuc	ACUUCACUAUUUUCAUCAA	-2.2	-0.9	4.6	0	2	1	-30.1	83	-1.6	8	II	26.3	75.1	92.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2354	NM_002559	Huesken	P2RX3	5024	CAUCAUGAUGGGCGUUUCCac	GGAAACGCCCAUCAUGAUG	-3.3	-2.1	-1.1	-1	1	5	-38.6	57.8	10.5	4	II	52.6	51.7	74.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2355	NM_002559	Huesken	P2RX3	5024	GUUUCCACUGUGUCCACCUcc	AGGUGGACACAGUGGAAAC	-2.1	-2.2	0.3	0	1	2	-40.1	65.8	-4	4	II	52.6	51.2	91.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2356	NM_002559	Huesken	P2RX3	5024	CCACCUCCGUGGGGCACCAgc	UGGUGCCCCACGGAGGUGG	-2.1	-3.3	-3.9	0	-2	5	-48.4	20.9	-5.8	-1	III	73.7	25.1	13.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2357	NM_002559	Huesken	P2RX3	5024	GCCCGCACUGGCUGUCUGAua	UCAGACAGCCAGUGCGGGC	-2.4	-3.4	-2.9	0	-3	6	-46.4	31.2	-6.1	0	III	68.4	22.4	23.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2358	NM_002559	Huesken	P2RX3	5024	ACACAGCGGUAUUUCUCCUca	AGGAGAAAUACCGCUGUGU	-2.1	-2.2	-0.4	3	3	4	-38.5	59.2	6.4	4	II	47.4	48.9	71.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2359	NM_002559	Huesken	P2RX3	5024	CACAGCGGUAUUUCUCCUCac	GAGGAGAAAUACCGCUGUG	-2.4	-2.1	-0.4	-1	0	4	-38.7	62.8	8.4	4	II	52.6	53.7	67.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2360	NM_002559	Huesken	P2RX3	5024	CCUCACUCUCUGGGCAGAAuc	UUCUGCCCAGAGAGUGAGG	-0.9	-3.3	-1	0	-3	4	-42.8	33	-3.1	1	III	57.9	25.1	44.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2361	NM_002559	Huesken	P2RX3	5024	UCACUCUCUGGGCAGAAUCcu	GAUUCUGCCCAGAGAGUGA	-2.4	-2.4	-2.4	2	2	4	-40.9	65.3	9.8	5	Ib	52.6	66.8	96.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2362	NM_002559	Huesken	P2RX3	5024	UCUCUGGGCAGAAUCCUUGca	CAAGGAUUCUGCCCAGAGA	-2.1	-2.4	-1.3	2	3	4	-40.5	62.9	-0.3	7	II	52.6	67.1	101.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2363	NM_002559	Huesken	P2RX3	5024	AGAAUCCUUGCAUCUGAUUuu	AAUCAGAUGCAAGGAUUCU	-0.9	-2.1	1	1	0	2	-34.9	71.8	-6.3	5	II	36.8	50.1	79.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2364	NM_002559	Huesken	P2RX3	5024	UCAGUAACAAUCAUCUUGGug	CCAAGAUGAUUGUUACUGA	-3.3	-2.4	0.6	0	4	2	-33.7	77.2	-2.6	8	Ia	36.8	75.2	98.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2365	NM_002559	Huesken	P2RX3	5024	AUGACAAAGACCGAGGUGCcc	GCACCUCGGUCUUUGUCAU	-3.4	-1.1	0.6	2	3	3	-39.6	58.3	2.7	7	Ia	52.6	65.3	85.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2366	NM_002559	Huesken	P2RX3	5024	GAGGUGCCCUGAGGUGGCGuc	CGCCACCUCAGGGCACCUC	-2.4	-2.4	-4.2	0	2	4	-47.5	26	0	0	II	73.7	36.1	13	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2367	NM_002559	Huesken	P2RX3	5024	GGCGUCACGUAAUCAGACAca	UGUCUGAUUACGUGACGCC	-2.1	-3.3	-1.9	1	-3	4	-39.2	48.6	-1.5	1	III	52.6	31.5	43.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2368	NM_002559	Huesken	P2RX3	5024	GUCACGUAAUCAGACACAUcc	AUGUGUCUGAUUACGUGAC	-1.1	-2.2	-1.9	1	-1	2	-35.5	55	-6.3	2	III	42.1	39.1	33.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2369	NM_002559	Huesken	P2RX3	5024	CAUGACUCUGUUGGCGUAGag	CUACGCCAACAGAGUCAUG	-2.1	-2.1	1.7	-1	0	4	-38.5	54.3	3.7	5	II	52.6	46.9	53.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2370	NM_002559	Huesken	P2RX3	5024	AAUGGCUGUGUCCCGUACCug	GGUACGGGACACAGCCAUU	-3.3	-0.9	-1.5	2	4	4	-41.9	59.9	7.5	3	Ib	57.9	66.4	61.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2371	NM_002559	Huesken	P2RX3	5024	UGCAAGAAAACCCACCCUAca	UAGGGUGGGUUUUCUUGCA	-1.3	-2.1	-2.6	0	-1	3	-38.8	55.8	-3.4	6	II	47.4	55.6	82.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2372	NM_002559	Huesken	P2RX3	5024	GAGAUGAUCAGAAGCUGAAcu	UUCAGCUUCUGAUCAUCUC	-0.9	-2.4	-1.2	0	-2	2	-36.5	56.4	-6.3	4	II	42.1	37.8	46.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2373	NM_002559	Huesken	P2RX3	5024	CUGAACUACUCGGUUGAUGau	CAUCAACCGAGUAGUUCAG	-2.1	-2.1	1.1	1	0	3	-36.3	63	0.7	4	II	47.4	50.8	74.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2374	NM_002559	Huesken	P2RX3	5024	CUCGGUUGAUGAUCCCGAUgg	AUCGGGAUCAUCAACCGAG	-1.1	-2.1	-4.7	0	-2	4	-39.4	52.9	-6	1	III	52.6	37.9	39.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2375	NM_002559	Huesken	P2RX3	5024	GGUUGAUGAUCCCGAUGGUcc	ACCAUCGGGAUCAUCAACC	-2.2	-3.3	-0.7	0	0	4	-40.1	43.9	-6.3	3	II	52.6	25.8	25.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2376	NM_002559	Huesken	P2RX3	5024	CUCUUCACAACCACCGACUug	AGUCGGUGGUUGUGAAGAG	-2.1	-2.1	1.3	-1	-2	3	-39.4	62.8	-13.3	5	II	52.6	51.8	64.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2377	NM_002559	Huesken	P2RX3	5024	CGGUGGACUGCUUCUCCGCug	GCGGAGAAGCAGUCCACCG	-3.4	-2.4	-2	-1	0	4	-44.5	45.8	6.1	0	II	68.4	46.4	81.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2378	NM_002559	Huesken	P2RX3	5024	GACUGCUUCUCCGCUGUGGuc	CCACAGCGGAGAAGCAGUC	-3.3	-2.4	-2.1	2	1	4	-43	54.6	-4.7	2	II	63.2	43.1	62.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2379	NM_002559	Huesken	P2RX3	5024	GCGAUUUUCAGCGUAGUCUca	AGACUACGCUGAAAAUCGC	-2.1	-3.4	0.5	2	-1	3	-36.7	69.9	-4	5	II	47.4	47.4	77.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2380	NM_002559	Huesken	P2RX3	5024	CAGCGUAGUCUCAUUCACCuc	GGUGAAUGAGACUACGCUG	-3.3	-2.1	-0.8	0	1	3	-38.8	55.3	2.8	2	II	52.6	59.3	87.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2381	NM_002559	Huesken	P2RX3	5024	AACUUCUUGGCUUUGUACUgg	AGUACAAAGCCAAGAAGUU	-2.1	-0.9	1.5	3	2	3	-34.1	93.2	-1	6	II	36.8	61.5	99.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2382	NM_002559	Huesken	P2RX3	5024	UUCUUGGCUUUGUACUGGUcg	ACCAGUACAAAGCCAAGAA	-2.2	-0.9	1.6	0	2	3	-36.5	72.9	-4	7	II	42.1	63.5	82.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2383	NM_002559	Huesken	P2RX3	5024	GGCUUUGUACUGGUCGGCCcc	GGCCGACCAGUACAAAGCC	-3.3	-3.3	-0.9	0	2	5	-42.9	51.8	9.7	2	II	63.2	44.7	68	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2384	NM_002559	Huesken	P2RX3	5024	UUGAGGAAGUUGAGCAGGAug	UCCUGCUCAACUUCCUCAA	-2.4	-0.9	3.6	0	1	2	-39.2	47.6	-6.3	6	II	47.4	57.5	90.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2385	NM_002559	Huesken	P2RX3	5024	UCCCACAGAAGUAAAGGCCgc	GGCCUUUACUUCUGUGGGA	-3.3	-2.4	-1	1	3	4	-40.3	59	10.5	5	II	52.6	65.5	80.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2386	NM_002559	Huesken	P2RX3	5024	CCACAGAAGUAAAGGCCGCca	GCGGCCUUUACUUCUGUGG	-3.4	-3.3	-0.3	0	1	6	-40.4	30.6	7.8	2	II	57.9	44.2	61.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2387	NM_002559	Huesken	P2RX3	5024	GAUGAUGGUGGGGAUGAUGuu	CAUCAUCCCCACCAUCAUC	-2.1	-2.4	4.8	0	1	4	-39.9	48	2.7	4	II	52.6	46	38.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2388	NM_002559	Huesken	P2RX3	5024	CUUCAGGAGGGUGCGGUACuc	GUACCGCACCCUCCUGAAG	-2.2	-2.1	1.7	1	0	4	-43.1	43.6	11.1	4	II	63.2	44.1	37.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2389	NM_002559	Huesken	P2RX3	5024	GGGACACGCUGCUUUUCUCag	GAGAAAAGCAGCGUGUCCC	-2.4	-3.3	1.4	1	1	3	-40.6	56.1	7.4	0	II	57.9	35.4	68.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2390	NM_002559	Huesken	P2RX3	5024	UCUCAGAAACGCUGUCGAGcc	CUCGACAGCGUUUCUGAGA	-2.1	-2.4	-1	0	3	3	-39	53	0.4	5	Ib	52.6	52.2	64.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2391	NM_002559	Huesken	P2RX3	5024	CAGAAACGCUGUCGAGCCGgg	CGGCUCGACAGCGUUUCUG	-2.4	-2.1	-1.7	-1	1	4	-41.2	40.7	6.1	1	II	63.2	40.8	86	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2392	NM_002559	Huesken	P2RX3	5024	AGAAACGCUGUCGAGCCGGgu	CCGGCUCGACAGCGUUUCU	-3.3	-2.1	-1.7	1	4	5	-42.4	39.3	0.4	4	Ib	63.2	58.2	98	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2393	NM_002559	Huesken	P2RX3	5024	AGGAGUAUUUGGGGAUGCAcu	UGCAUCCCCAAAUACUCCU	-2.1	-2.1	2.3	0	0	4	-39.4	49.5	-3.8	4	II	47.4	41.7	56.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2394	NM_002559	Huesken	P2RX3	5024	AAGUCGCACACCCAGCCGAuc	UCGGCUGGGUGUGCGACUU	-2.4	-0.9	-1.2	2	1	4	-44.3	46.1	-6.1	3	II	63.2	47.4	77.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2395	NM_002559	Huesken	P2RX3	5024	CACACCCAGCCGAUCUUAAug	UUAAGAUCGGCUGGGUGUG	-0.9	-2.1	0.5	0	-3	4	-39.6	53.7	-8.4	3	III	52.6	29.6	55.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2396	NM_002559	Huesken	P2RX3	5024	CCAGCCGAUCUUAAUGCCCag	GGGCAUUAAGAUCGGCUGG	-3.3	-3.3	-1	-1	1	4	-40.9	45.6	5.8	0	II	57.9	51.5	59.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2397	NM_002559	Huesken	P2RX3	5024	UGCGCGCCAGUUUGGCAAAau	UUUGCCAAACUGGCGCGCA	-0.9	-2.1	-4.5	-1	-2	7	-41.3	46.4	-0.7	1	II	57.9	37	37.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2398	NM_002559	Huesken	P2RX3	5024	UGGCAAAAUCCUGCCCCGCaa	GCGGGGCAGGAUUUUGCCA	-3.4	-2.1	-5.5	1	3	7	-43.7	44.3	10.8	3	Ib	63.2	60.8	67.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2399	NM_002559	Huesken	P2RX3	5024	GCAAAAUCCUGCCCCGCAAac	UUGCGGGGCAGGAUUUUGC	-0.9	-3.4	-3	2	-2	7	-41.3	42.6	-1	2	II	57.9	27.7	33.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2400	NM_002559	Huesken	P2RX3	5024	GACCACGUCCCCUACCCGCaa	GCGGGUAGGGGACGUGGUC	-3.4	-2.4	-1.1	2	2	5	-47.1	38.5	5.4	1	II	73.7	45.8	37.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2401	NM_002559	Huesken	P2RX3	5024	ACGUCCCCUACCCGCAAGAug	UCUUGCGGGUAGGGGACGU	-2.4	-2.2	-1.4	0	-1	5	-44.6	43.4	-6.1	2	II	63.2	36.9	75.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2402	NM_002559	Huesken	P2RX3	5024	CCGCAAGAUGGGGCAGAAAgg	UUUCUGCCCCAUCUUGCGG	-0.9	-3.3	-2.2	0	-5	5	-41.5	24.2	-3.1	0	III	57.9	20.4	35.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2403	NM_002559	Huesken	P2RX3	5024	GGUGGAAGCGGCAGGUCUUca	AAGACCUGCCGCUUCCACC	-0.9	-3.3	1.7	0	-1	5	-43.9	29	-8.9	1	III	63.2	28.3	39.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2404	NM_002559	Huesken	P2RX3	5024	UCUUCAUGUCCCUGGCUGUca	ACAGCCAGGGACAUGAAGA	-2.2	-2.4	0.8	0	1	3	-41.6	64.1	-8.7	7	II	52.6	51.9	78.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2405	NM_002559	Huesken	P2RX3	5024	UUCAUGUCCCUGGCUGUCAgg	UGACAGCCAGGGACAUGAA	-2.1	-0.9	0.8	2	0	3	-41.6	66	-3.8	8	II	52.6	63.1	90.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2406	NM_002559	Huesken	P2RX3	5024	CAUGUCCCUGGCUGUCAGGuu	CCUGACAGCCAGGGACAUG	-3.3	-2.1	-4.7	-2	1	3	-43.7	50	-4.4	1	II	63.2	44	33.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2407	NM_002559	Huesken	P2RX3	5024	UGGCUGUCAGGUUGGGAAGga	CUUCCCAACCUGACAGCCA	-2.1	-2.1	2.4	1	1	3	-42.3	59.7	3	4	II	57.9	54.1	66.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2408	NM_002559	Huesken	P2RX3	5024	UUCUUGAUGAAAAUAGUGAag	UCACUAUUUUCAUCAAGAA	-2.4	-0.9	1.1	1	1	1	-30.3	89.4	-1.7	9	II	26.3	68.9	79.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2409	NM_002559	Huesken	P2RX3	5024	UAGUGAAGUUCUCAGCUUCca	GAAGCUGAGAACUUCACUA	-2.4	-1.3	-2.3	0	2	2	-36.1	73.5	10.5	6	Ia	42.1	68.1	88.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2410	NM_002559	Huesken	P2RX3	5024	CCAUCAUGAUGGGCGUUUCca	GAAACGCCCAUCAUGAUGG	-2.4	-3.3	-2.4	0	-1	5	-38.6	57.2	5.4	3	II	52.6	41	48.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2411	NM_002559	Huesken	P2RX3	5024	AGCCCUGGAUCUCACAGGUcc	ACCUGUGAGAUCCAGGGCU	-2.2	-2.1	-4.4	2	1	4	-44	49.1	-0.9	1	II	57.9	42.5	50.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2412	NM_002559	Huesken	P2RX3	5024	ACAGGUCCGGAGCACAGAGcu	CUCUGUGCUCCGGACCUGU	-2.1	-2.2	-1.4	2	3	4	-44.2	38.8	4.7	1	II	63.2	47.8	41.9	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2413	NM_002559	Huesken	P2RX3	5024	GUCCGGAGCACAGAGCUGUag	ACAGCUCUGUGCUCCGGAC	-2.2	-2.2	-1.8	2	-1	4	-44.3	33.2	-6.3	2	III	63.2	40.7	59.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2414	NM_002559	Huesken	P2RX3	5024	CACAGAGCUGUAGUUCACGca	CGUGAACUACAGCUCUGUG	-2.4	-2.1	-0.5	-1	1	2	-38.4	49.1	0.3	3	II	52.6	52.5	42	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2415	NM_002559	Huesken	P2RX3	5024	UGUAGUUCACGCAGCGGCCag	GGCCGCUGCGUGAACUACA	-3.3	-2.1	-0.3	0	4	6	-43.3	56.8	2.5	5	Ia	63.2	65.7	72.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2416	NM_002559	Huesken	P2RX3	5024	ACGCAGCGGCCAGUGAGGAuc	UCCUCACUGGCCGCUGCGU	-2.4	-2.2	-0.3	3	1	6	-46.7	21	-1.1	2	II	68.4	32.4	16.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2417	NM_002559	Huesken	P2RX3	5024	CCUUGCAUCUGAUUUUCAGua	CUGAAAAUCAGAUGCAAGG	-2.1	-3.3	-1.8	0	1	2	-34.4	73.9	0.3	2	II	42.1	43.1	85.5	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2418	NM_002559	Huesken	P2RX3	5024	AUCUGAUUUUCAGUAACAAuc	UUGUUACUGAAAAUCAGAU	-0.9	-1.1	-2.3	2	1	1	-30.1	73.2	-3.8	4	II	26.3	47.9	54.1	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2419	NM_002559	Huesken	P2RX3	5024	UCUGAUUUUCAGUAACAAUca	AUUGUUACUGAAAAUCAGA	-1.1	-2.4	-2.3	1	0	1	-30.1	84.3	1	5	II	26.3	60.2	74.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2420	NM_002559	Huesken	P2RX3	5024	UGAUGAUGACAAAGACCGAgg	UCGGUCUUUGUCAUCAUCA	-2.4	-2.1	0.9	0	1	3	-36.6	52.8	-4	6	II	42.1	50.9	72.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2421	NM_002559	Huesken	P2RX3	5024	CCGAGGUGCCCUGAGGUGGcg	CCACCUCAGGGCACCUCGG	-3.3	-3.3	-0.7	1	-1	4	-47.4	33	-1.3	2	II	73.7	41.9	70.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2422	NM_002559	Huesken	P2RX3	5024	AGGUGCCCUGAGGUGGCGUca	ACGCCACCUCAGGGCACCU	-2.2	-2.1	-4.2	1	2	4	-47.3	40.5	0.7	1	II	68.4	39.8	8.6	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2423	NM_002559	Huesken	P2RX3	5024	GGUGGCGUCACGUAAUCAGac	CUGAUUACGUGACGCCACC	-2.1	-3.3	-1.3	1	1	4	-40.1	48.4	-2.4	1	II	57.9	39.1	50	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2424	NM_002559	Huesken	P2RX3	5024	UCCAUGACUCUGUUGGCGUag	ACGCCAACAGAGUCAUGGA	-2.2	-2.4	-0.6	0	2	4	-40.8	64.2	-1.6	6	II	52.6	59.1	89.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2425	NM_002559	Huesken	P2RX3	5024	UUGGCGUAGAGUCCGGAGCcc	GCUCCGGACUCUACGCCAA	-3.4	-0.9	-1	2	3	4	-43.7	63.8	12.8	4	II	63.2	70.9	87.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2426	NM_002559	Huesken	P2RX3	5024	GAGCCCUUCACCUUGGUUAcc	UAACCAAGGUGAAGGGCUC	-1.3	-2.4	-1	3	-2	4	-40.3	67.7	-6.1	3	III	52.6	40.5	16.2	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2427	NM_002559	Huesken	P2RX3	5024	AAAACCCACCCUACAAAGUag	ACUUUGUAGGGUGGGUUUU	-2.2	-0.9	0	2	2	3	-36.2	67.4	-5.9	5	II	42.1	55.4	89.4	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2428	NM_002559	Huesken	P2RX3	5024	GAUGAUCAGAAGCUGAACUac	AGUUCAGCUUCUGAUCAUC	-2.1	-2.4	-1.2	1	1	2	-36.3	67.9	5.8	3	II	42.1	48.1	72.3	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2429	NM_002559	Huesken	P2RX3	5024	AUCAGAAGCUGAACUACUCgg	GAGUAGUUCAGCUUCUGAU	-2.4	-1.1	-1.2	2	3	2	-36.3	63.8	4.8	4	Ia	42.1	62.3	91.8	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2430	NM_002559	Huesken	P2RX3	5024	GAAGCUGAACUACUCGGUUga	AACCGAGUAGUUCAGCUUC	-0.9	-2.4	-0.9	0	1	3	-37.4	53.6	0.8	3	III	47.4	43	62	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2431	NM_002559	Huesken	P2RX3	5024	GAUGAUCCCGAUGGUCCAGcu	CUGGACCAUCGGGAUCAUC	-2.1	-2.4	-0.7	0	2	4	-41.5	47.4	8	2	II	57.9	38.3	60.7	Luciferase reporter assay	HeLa	50nM	48h	"Design of a genome-wide siRNA library using an artificial
neural network"	16025102	2005	The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.	Nature biotechnology	2005/8/1
Si2432	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2433	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2434	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2435	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2436	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2437	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2438	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2439	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2440	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2441	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2442	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2443	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2444	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2445	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2446	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2447	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2448	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2449	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2450	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2451	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2452	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2453	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2454	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2455	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2456	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2457	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2458	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2459	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2460	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2461	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2462	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2463	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2464	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2465	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2466	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2467	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2468	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2469	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2470	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2471	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2472	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2473	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2474	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2475	AH001498	Harborth	Lamin A	33782	GAGCUCCUGCAGGUCCUCCuu	GGAGGACCUGCAGGAGCUC	-3.3	-2.4	-1	2	2	2	-46.5	48	12	1	II	68.4	46	83	Western blotting	HeLa	100nM	24h	Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing.	12804036	2003	Small interfering RNAs (siRNAs) induce sequence-specific gene silencing in mammalian cells and guide mRNA degradation in the process of RNA interference (RNAi). By targeting endogenous lamin A/C mRNA in human HeLa or mouse SW3T3 cells, we investigated the positional variation of siRNA-mediated gene silencing. We find cell-type-dependent global effects and cell-type-independent positional effects. HeLa cells were about 2-fold more responsive to siRNAs than SW3T3 cells but displayed a very similar pattern of positional variation of lamin A/C silencing. In HeLa cells, 26 of 44 tested standard 21-nucleotide (nt) siRNA duplexes reduced the protein expression by at least 90%, and only 2 duplexes reduced the lamin A/C proteins to <50%. Fluorescent chromophores did not perturb gene silencing when conjugated to the 5'-end or 3'-end of the sense siRNA strand and the 5'-end of the antisense siRNA strand, but conjugation to the 3'-end of the antisense siRNA abolished gene silencing. RNase-protecting phosphorothioate and 2'-fluoropyrimidine RNA backbone modifications of siRNAs did not significantly affect silencing efficiency, although cytotoxic effects were observed when every second phosphate of an siRNA duplex was replaced by phosphorothioate. Synthetic RNA hairpin loops were subsequently evaluated for lamin A/C silencing as a function of stem length and loop composition. As long as the 5'-end of the guide strand coincided with the 5'-end of the hairpin RNA, 19-29 base pair (bp) hairpins effectively silenced lamin A/C, but when the hairpin started with the 5'-end of the sense strand, only 21-29 bp hairpins were highly active.	Antisense & Nucleic Acid Drug Development	2003/4/1
Si2476	M15077	Ui-Tei	FireflyLuc	116160065	UCUUUAUGUUUUUGGCGUCuu	GACGCCAAAAACAUAAAGA	-2.4	-2.4		0	3	4	-32.4	80.6	10.4	7	Ia	36.8	62.9	85.8	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2477	M15077	Ui-Tei	FireflyLuc	116160065	UUCUUUAUGUUUUUGGCGUcu	ACGCCAAAAACAUAAAGAA	-2.2	-0.9		1	3	4	-30.9	97.4	1.4	7	II	31.6	70.3	94.9	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2478	M15077	Ui-Tei	FireflyLuc	116160065	GGGCCUUUCUUUAUGUUUUug	AAAACAUAAAGAAAGGCCC	-0.9	-3.3		3	-2	5	-32.9	77.2	-1.2	2	III	36.8	41.3	11.1	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2479	M15077	Ui-Tei	FireflyLuc	116160065	CCGGGCCUUUCUUUAUGUUuu	AACAUAAAGAAAGGCCCGG	-0.9	-3.3		-1	-2	7	-36.8	55.4	-3	0	III	47.4	29.8	7.6	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2480	M15077	Ui-Tei	FireflyLuc	116160065	GGCGCCGGGCCUUUCUUUAug	UAAAGAAAGGCCCGGCGCC	-1.3	-3.3	-2.2	1	-3	11	-43	46.4	-0.8	0	III	63.2	23.8	5.9	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2481	M15077	Ui-Tei	FireflyLuc	116160065	UAGAAUGGCGCCGGGCCUUuc	AAGGCCCGGCGCCAUUCUA	-0.9	-1.3	-3.9	0	1	11	-44.4	41.4	-6	5	II	63.2	55.6	23	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2482	M15077	Ui-Tei	FireflyLuc	116160065	AGCGGAUAGAAUGGCGCCGgg	CGGCGCCAUUCUAUCCGCU	-2.4	-2.1	-1.9	2	3	7	-42.8	41.2	5.3	2	II	63.2	52.3	53.8	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2483	M15077	Ui-Tei	FireflyLuc	116160065	UCUUCCAGCGGAUAGAAUGgc	CAUUCUAUCCGCUGGAAGA	-2.1	-2.4	-1.7	1	2	4	-37.8	73.2	0	5	Ib	47.4	64.9	93.2	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2484	M15077	Ui-Tei	FireflyLuc	116160065	GUUCCAUCUUCCAGCGGAUag	AUCCGCUGGAAGAUGGAAC	-1.1	-2.2	-2.2	1	0	4	-39.9	46.4	-11.3	2	II	52.6	32.1	23.4	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2485	M15077	Ui-Tei	FireflyLuc	116160065	CCAGCGGUUCCAUCUUCCAgc	UGGAAGAUGGAACCGCUGG	-2.1	-3.3	0.3	-1	-2	4	-41.8	42.4	-8.1	1	III	57.9	36.8	16.6	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2486	M15077	Ui-Tei	FireflyLuc	116160065	UGCUCUCCAGCGGUUCCAUcu	AUGGAACCGCUGGAGAGCA	-1.1	-2.1	-2.6	0	1	4	-43.1	57.6	-4	2	II	57.9	43.6	35.4	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2487	M15077	Ui-Tei	FireflyLuc	116160065	UUUCAUAGCUUCUGCCAACcg	GUUGGCAGAAGCUAUGAAA	-2.2	-0.9	-2.7	1	3	3	-35.7	71.7	7.8	6	Ia	42.1	69.1	93.1	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2488	M15077	Ui-Tei	FireflyLuc	116160065	CGCGCCCAACACCGGCAUAaa	UAUGCCGGUGUUGGGCGCG	-1.3	-2.4	-4.1	-2	-5	7	-44.6	28.3	-3.4	-2	III	68.4	17.1	19.7	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2489	M15077	Ui-Tei	FireflyLuc	116160065	UAAAUAACGCGCCCAACACcg	GUGUUGGGCGCGUUAUUUA	-2.2	-1.3	-2.2	1	4	7	-36.3	69.2	7.5	7	Ia	47.4	69.4	91.9	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2490	M15077	Ui-Tei	FireflyLuc	116160065	CGCGGGCGCAACUGCAACUcc	AGUUGCAGUUGCGCCCGCG	-2.1	-2.4	-1.5	-1	-3	9	-44.1	32.4	-6	-1	III	68.4	33.9	37.3	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2491	M15077	Ui-Tei	FireflyLuc	116160065	UUAUAAAUGUCGUUCGCGGgc	CCGCGAACGACAUUUAUAA	-3.3	-0.9	0.7	1	6	5	-33.6	78.2	2.7	8	Ia	42.1	73.9	94.9	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2492	M15077	Ui-Tei	FireflyLuc	116160065	GGGAGCUUUUUUUGCACGUuc	ACGUGCAAAAAAAGCUCCC	-2.2	-3.3	-0.2	0	-1	3	-36.4	58.9	-1	2	III	47.4	38	30	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2493	M15077	Ui-Tei	FireflyLuc	116160065	UAAUUUUUUGGAUGAUUGGga	CCAAUCAUCCAAAAAAUUA	-3.3	-1.3	3.8	0	4	2	-28.6	98.5	0	9	Ia	26.3	77.9	97.5	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2494	M15077	Ui-Tei	FireflyLuc	116160065	AUUCAUUAAAACCGGGAGGua	CCUCCCGGUUUUAAUGAAU	-3.3	-1.1	0.3	2	4	5	-34.8	72.9	5.1	7	Ia	42.1	65.6	73.5	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2495	M15077	Ui-Tei	FireflyLuc	116160065	UUCUAUGAGGCAGAGCGACac	GUCGCUCUGCCUCAUAGAA	-2.2	-0.9	-0.4	1	3	3	-40.2	54.8	7.8	7	Ia	52.6	63.4	69.9	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2496	M15077	Ui-Tei	FireflyLuc	116160065	AAUAGGAUCUCUGGCAUGCga	GCAUGCCAGAGAUCCUAUU	-3.4	-0.9	1.5	1	4	3	-38.7	65.5	8.1	7	Ia	47.4	63.8	94.5	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2497	M15077	Ui-Tei	FireflyLuc	116160065	UUAAAAUCGCAGUAUCCGGaa	CCGGAUACUGCGAUUUUAA	-3.3	-0.9	0.7	2	5	4	-34.4	76.1	5	8	Ia	42.1	76.1	96.3	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2498	M15077	Ui-Tei	FireflyLuc	116160065	UAGUAAACAUUCCAAAACCgu	GGUUUUGGAAUGUUUACUA	-3.3	-1.3	1.4	0	3	2	-31	79	10.2	7	Ia	31.6	76.7	93.3	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2499	M15077	Ui-Tei	FireflyLuc	116160065	AUUAAGACGACUCGAAAUCca	GAUUUCGAGUCGUCUUAAU	-2.4	-1.1	-0.5	1	3	2	-32.5	72.9	13.2	8	Ia	36.8	66.9	98.9	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2500	M15077	Ui-Tei	FireflyLuc	116160065	AAUCUUGUAAUCCUGAAGGcu	CCUUCAGGAUUACAAGAUU	-3.3	-0.9	-0.5	2	5	2	-33.6	85.2	2.4	7	Ia	36.8	72	98.9	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2501	M15077	Ui-Tei	FireflyLuc	116160065	UAGAUAAAUCGUAUUUGUCaa	GACAAAUACGAUUUAUCUA	-2.4	-1.3	1.4	0	4	2	-29	94	10.2	7	Ia	26.3	74.8	86.5	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2502	M15077	Ui-Tei	FireflyLuc	116160065	AUUAGAUAAAUCGUAUUUGuc	CAAAUACGAUUUAUCUAAU	-2.1	-1.1	1.4	2	4	2	-26.4	89.6	0	8	Ia	21.1	70.7	87.8	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2503	M15077	Ui-Tei	FireflyLuc	116160065	CCCUCGGGUGUAAUCAGAAua	UUCUGAUUACACCCGAGGG	-0.9	-3.3	-1.1	-1	-3	4	-39.8	41.3	-6	1	III	52.6	29.8	10.7	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2504	M15077	Ui-Tei	FireflyLuc	116160065	CCGCGCCCGGUUUAUCAUCcc	GAUGAUAAACCGGGCGCGG	-2.4	-3.3	-2	1	-2	10	-41.6	40	8.4	-2	II	63.2	38.8	52.9	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2505	M15077	Ui-Tei	FireflyLuc	116160065	AACUUUACCGACCGCGCCCgg	GGGCGCGGUCGGUAAAGUU	-3.3	-0.9	-1.3	1	4	8	-42.3	60.8	9.8	4	Ia	63.2	64.9	93.6	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2506	M15077	Ui-Tei	FireflyLuc	116160065	AAAAUGGAACAACUUUACCga	GGUAAAGUUGUUCCAUUUU	-3.3	-0.9	0	1	5	2	-30.6	83.6	14.4	7	Ia	31.6	72.3	94.5	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2507	M15077	Ui-Tei	FireflyLuc	116160065	UACAUAACCGGACAUAAUCau	GAUUAUGUCCGGUUAUGUA	-2.4	-1.3	0.9	1	2	4	-33.6	80.3	9.8	7	Ia	36.8	73.5	96.8	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2508	M15077	Ui-Tei	FireflyLuc	116160065	AGAAUGUAGCCAUCCAUCCuu	GGAUGGAUGGCUACAUUCU	-3.3	-2.1	-4	3	3	3	-38.7	69.2	5.1	6	Ia	47.4	68.9	97.4	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2509	M15077	Ui-Tei	FireflyLuc	116160065	UUAAUCAGAGACUUCAGGCgg	GCCUGAAGUCUCUGAUUAA	-3.4	-0.9	0.8	0	6	3	-36.2	86.6	9.7	8	Ia	42.1	81.5	95.6	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2510	M15077	Ui-Tei	FireflyLuc	116160065	CGGCGGCGGGAAGUUCACCgg	GGUGAACUUCCCGCCGCCG	-3.3	-2.4	-4.5	1	0	10	-45.7	24.8	10	-2	II	73.7	39.1	14.5	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2511	M15077	Ui-Tei	FireflyLuc	116160065	AAACAACAACGGCGGCGGGaa	CCCGCCGCCGUUGUUGUUU	-3.3	-0.9	1.6	1	5	10	-41.6	36.2	5.1	4	Ia	63.2	46.9	42.3	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2512	M15077	Ui-Tei	FireflyLuc	116160065	CAAAACAACAACGGCGGCGgg	CGCCGCCGUUGUUGUUUUG	-2.4	-2.1	1.9	-1	2	8	-38	48.7	0.7	5	II	57.9	56.5	51.3	Luciferase reporter assay	CHO-K1;HeLa;E14TG2a ;S2	50nM	24h	Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference	14769950	2004	In the present study, the relationship between short interfering RNA (siRNA) sequence and RNA interference (RNAi) effect was extensively analyzed using 62 targets of four exogenous and two endogenous genes and three mammalian and Drosophila cells. We present the rules that may govern siRNA sequence preference and in accordance with which highly effective siRNAs essential for systematic mammalian functional genomics can be readily designed. These rules indicate that siRNAs which simultaneously satisfy all four of the following sequence conditions are capable of inducing highly effective gene silencing in mammalian cells: (i) A/U at the 5' end of the antisense strand; (ii) G/C at the 5' end of the sense strand; (iii) at least five A/U residues in the 5' terminal one-third of the antisense strand; and (iv) the absence of any GC stretch of more than 9 nt in length. siRNAs opposite in features with respect to the first three conditions give rise to little or no gene silencing in mammalian cells. Essentially the same rules for siRNA sequence preference were found applicable to DNA-based RNAi in mammalian cells and in ovo RNAi using chick embryos. In contrast to mammalian and chick cells, little siRNA sequence preference could be detected in Drosophila in vivo RNAi.	Nucleic Acids Research	2004/3/1
Si2513	NM_001632	Khovorova	SEAP	250	CAGAAGUCCGGGUUCUCCUcc	AGGAGAACCCGGACUUCUG	-2.1	-2.1	-2.7	1	0	5	-41.9	60.3	-0.9	2	II	57.9	51.8	81	Flow cytometry and immunofluorescence	HEK293	0.25ug/Ml	24h	Functional siRNAs and miRNAs Exhibit Strand Biasunctional siRNAs and miRNAs Exhibit Strand Bias	14567918	2003	Both microRNAs (miRNA) and small interfering RNAs (siRNA) share a common set of cellular proteins (Dicer and the RNA-induced silencing complex [RISC]) to elicit RNA interference. In the following work, a statistical analysis of the internal stability of published miRNA sequences in the context of miRNA precursor hairpins revealed enhanced flexibility of miRNA precursors, especially at the 5′-anti-sense (AS) terminal base pair. The same trend was observed in siRNA, with functional duplexes displaying a lower internal stability (Δ0.5 kcal/mol) at the 5′-AS end than nonfunctional duplexes. Average internal stability of siRNA molecules retrieved from plant cells after introduction of long RNA sequences also shows this characteristic thermodynamic signature. Together, these results suggest that the thermodynamic properties of siRNA play a critical role in determining the molecule's function and longevity, possibly biasing the steps involved in duplex unwinding and strand retention by RISC.	Cell	2003/10/17
Si2514	NM_001632	Khovorova	SEAP	250	CCCAGGAAGAUGAUGAGGUuc	ACCUCAUCAUCUUCCUGGG	-2.2	-3.3	-0.3	0	-1	3	-40.7	44.7	-6.2	3	III	52.6	42.4	49	Flow cytometry and immunofluorescence	HEK293	0.25ug/Ml	24h	Functional siRNAs and miRNAs Exhibit Strand Biasunctional siRNAs and miRNAs Exhibit Strand Bias	14567918	2003	Both microRNAs (miRNA) and small interfering RNAs (siRNA) share a common set of cellular proteins (Dicer and the RNA-induced silencing complex [RISC]) to elicit RNA interference. In the following work, a statistical analysis of the internal stability of published miRNA sequences in the context of miRNA precursor hairpins revealed enhanced flexibility of miRNA precursors, especially at the 5′-anti-sense (AS) terminal base pair. The same trend was observed in siRNA, with functional duplexes displaying a lower internal stability (Δ0.5 kcal/mol) at the 5′-AS end than nonfunctional duplexes. Average internal stability of siRNA molecules retrieved from plant cells after introduction of long RNA sequences also shows this characteristic thermodynamic signature. Together, these results suggest that the thermodynamic properties of siRNA play a critical role in determining the molecule's function and longevity, possibly biasing the steps involved in duplex unwinding and strand retention by RISC.	Cell	2003/10/17
Si2515	NM_001632	Khovorova	SEAP	250	ACUGUCUGGCACAUGUUUGuc	CAAACAUGUGCCAGACAGU	-2.1	-2.2	0	3	3	3	-37.6	70.2	-0.3	6	Ib	47.4	57.6	88	Flow cytometry and immunofluorescence	HEK293	0.25ug/Ml	24h	Functional siRNAs and miRNAs Exhibit Strand Biasunctional siRNAs and miRNAs Exhibit Strand Bias	14567918	2003	Both microRNAs (miRNA) and small interfering RNAs (siRNA) share a common set of cellular proteins (Dicer and the RNA-induced silencing complex [RISC]) to elicit RNA interference. In the following work, a statistical analysis of the internal stability of published miRNA sequences in the context of miRNA precursor hairpins revealed enhanced flexibility of miRNA precursors, especially at the 5′-anti-sense (AS) terminal base pair. The same trend was observed in siRNA, with functional duplexes displaying a lower internal stability (Δ0.5 kcal/mol) at the 5′-AS end than nonfunctional duplexes. Average internal stability of siRNA molecules retrieved from plant cells after introduction of long RNA sequences also shows this characteristic thermodynamic signature. Together, these results suggest that the thermodynamic properties of siRNA play a critical role in determining the molecule's function and longevity, possibly biasing the steps involved in duplex unwinding and strand retention by RISC.	Cell	2003/10/17
Si2516	NM_001632	Khovorova	SEAP	250	GGCGAGGCGUGCUGCACUCgu	GAGUGCAGCACGCCUCGCC	-2.4	-3.3	-0.7	0	0	4	-46.8	25.7	7.8	1	II	73.7	32.6	-5	Flow cytometry and immunofluorescence	HEK293	0.25ug/Ml	24h	Functional siRNAs and miRNAs Exhibit Strand Biasunctional siRNAs and miRNAs Exhibit Strand Bias	14567918	2003	Both microRNAs (miRNA) and small interfering RNAs (siRNA) share a common set of cellular proteins (Dicer and the RNA-induced silencing complex [RISC]) to elicit RNA interference. In the following work, a statistical analysis of the internal stability of published miRNA sequences in the context of miRNA precursor hairpins revealed enhanced flexibility of miRNA precursors, especially at the 5′-anti-sense (AS) terminal base pair. The same trend was observed in siRNA, with functional duplexes displaying a lower internal stability (Δ0.5 kcal/mol) at the 5′-AS end than nonfunctional duplexes. Average internal stability of siRNA molecules retrieved from plant cells after introduction of long RNA sequences also shows this characteristic thermodynamic signature. Together, these results suggest that the thermodynamic properties of siRNA play a critical role in determining the molecule's function and longevity, possibly biasing the steps involved in duplex unwinding and strand retention by RISC.	Cell	2003/10/17
Si2517	NM_001632	Khovorova	SEAP	250	CAGCCAUUCCUGCACCAGAuu	UCUGGUGCAGGAAUGGCUG	-2.4	-2.1	-0.4	2	-3	3	-42.5	45.9	-3.4	1	III	57.9	40.1	30	Flow cytometry and immunofluorescence	HEK293	0.25ug/Ml	24h	Functional siRNAs and miRNAs Exhibit Strand Biasunctional siRNAs and miRNAs Exhibit Strand Bias	14567918	2003	Both microRNAs (miRNA) and small interfering RNAs (siRNA) share a common set of cellular proteins (Dicer and the RNA-induced silencing complex [RISC]) to elicit RNA interference. In the following work, a statistical analysis of the internal stability of published miRNA sequences in the context of miRNA precursor hairpins revealed enhanced flexibility of miRNA precursors, especially at the 5′-anti-sense (AS) terminal base pair. The same trend was observed in siRNA, with functional duplexes displaying a lower internal stability (Δ0.5 kcal/mol) at the 5′-AS end than nonfunctional duplexes. Average internal stability of siRNA molecules retrieved from plant cells after introduction of long RNA sequences also shows this characteristic thermodynamic signature. Together, these results suggest that the thermodynamic properties of siRNA play a critical role in determining the molecule's function and longevity, possibly biasing the steps involved in duplex unwinding and strand retention by RISC.	Cell	2003/10/17
Si2518	NM_001632	Khovorova	SEAP	250	GAACAUGAUCGUCUCAGUCag	GACUGAGACGAUCAUGUUC	-2.4	-2.4	-0.2	1	3	2	-37	64	17.9	4	II	47.4	52.6	85	Flow cytometry and immunofluorescence	HEK293	0.25ug/Ml	24h	Functional siRNAs and miRNAs Exhibit Strand Biasunctional siRNAs and miRNAs Exhibit Strand Bias	14567918	2003	Both microRNAs (miRNA) and small interfering RNAs (siRNA) share a common set of cellular proteins (Dicer and the RNA-induced silencing complex [RISC]) to elicit RNA interference. In the following work, a statistical analysis of the internal stability of published miRNA sequences in the context of miRNA precursor hairpins revealed enhanced flexibility of miRNA precursors, especially at the 5′-anti-sense (AS) terminal base pair. The same trend was observed in siRNA, with functional duplexes displaying a lower internal stability (Δ0.5 kcal/mol) at the 5′-AS end than nonfunctional duplexes. Average internal stability of siRNA molecules retrieved from plant cells after introduction of long RNA sequences also shows this characteristic thermodynamic signature. Together, these results suggest that the thermodynamic properties of siRNA play a critical role in determining the molecule's function and longevity, possibly biasing the steps involved in duplex unwinding and strand retention by RISC.	Cell	2003/10/17
Si2519	NM_001632	Khovorova	SEAP	250	CUGGUGAGCUGGCCCGCCCuc	GGGCGGGCCAGCUCACCAG	-3.3	-2.1	-5.5	0	1	9	-49.6	35.3	7.8	1	II	78.9	51.9	20	Flow cytometry and immunofluorescence	HEK293	0.25ug/Ml	24h	Functional siRNAs and miRNAs Exhibit Strand Biasunctional siRNAs and miRNAs Exhibit Strand Bias	14567918	2003	Both microRNAs (miRNA) and small interfering RNAs (siRNA) share a common set of cellular proteins (Dicer and the RNA-induced silencing complex [RISC]) to elicit RNA interference. In the following work, a statistical analysis of the internal stability of published miRNA sequences in the context of miRNA precursor hairpins revealed enhanced flexibility of miRNA precursors, especially at the 5′-anti-sense (AS) terminal base pair. The same trend was observed in siRNA, with functional duplexes displaying a lower internal stability (Δ0.5 kcal/mol) at the 5′-AS end than nonfunctional duplexes. Average internal stability of siRNA molecules retrieved from plant cells after introduction of long RNA sequences also shows this characteristic thermodynamic signature. Together, these results suggest that the thermodynamic properties of siRNA play a critical role in determining the molecule's function and longevity, possibly biasing the steps involved in duplex unwinding and strand retention by RISC.	Cell	2003/10/17
Si2520	NM_001632	Khovorova	SEAP	250	GUGGGAGUGGUCGGCAGUGac	CACUGCCGACCACUCCCAC	-2.1	-2.2	1.9	0	1	4	-45.2	21.8	-2.4	1	II	68.4	39.2	-10	Flow cytometry and immunofluorescence	HEK293	0.25ug/Ml	24h	Functional siRNAs and miRNAs Exhibit Strand Biasunctional siRNAs and miRNAs Exhibit Strand Bias	14567918	2003	Both microRNAs (miRNA) and small interfering RNAs (siRNA) share a common set of cellular proteins (Dicer and the RNA-induced silencing complex [RISC]) to elicit RNA interference. In the following work, a statistical analysis of the internal stability of published miRNA sequences in the context of miRNA precursor hairpins revealed enhanced flexibility of miRNA precursors, especially at the 5′-anti-sense (AS) terminal base pair. The same trend was observed in siRNA, with functional duplexes displaying a lower internal stability (Δ0.5 kcal/mol) at the 5′-AS end than nonfunctional duplexes. Average internal stability of siRNA molecules retrieved from plant cells after introduction of long RNA sequences also shows this characteristic thermodynamic signature. Together, these results suggest that the thermodynamic properties of siRNA play a critical role in determining the molecule's function and longevity, possibly biasing the steps involved in duplex unwinding and strand retention by RISC.	Cell	2003/10/17
Si2521	NM_001632	Khovorova	SEAP	250	AACAUCCGGCCGGGCGCCGuc	CGGCGCCCGGCCGGAUGUU	-2.4	-0.9	-6.4	1	4	14	-48.3	29.8	-2.4	3	Ib	78.9	48	60	Flow cytometry and immunofluorescence	HEK293	0.25ug/Ml	24h	Functional siRNAs and miRNAs Exhibit Strand Biasunctional siRNAs and miRNAs Exhibit Strand Bias	14567918	2003	Both microRNAs (miRNA) and small interfering RNAs (siRNA) share a common set of cellular proteins (Dicer and the RNA-induced silencing complex [RISC]) to elicit RNA interference. In the following work, a statistical analysis of the internal stability of published miRNA sequences in the context of miRNA precursor hairpins revealed enhanced flexibility of miRNA precursors, especially at the 5′-anti-sense (AS) terminal base pair. The same trend was observed in siRNA, with functional duplexes displaying a lower internal stability (Δ0.5 kcal/mol) at the 5′-AS end than nonfunctional duplexes. Average internal stability of siRNA molecules retrieved from plant cells after introduction of long RNA sequences also shows this characteristic thermodynamic signature. Together, these results suggest that the thermodynamic properties of siRNA play a critical role in determining the molecule's function and longevity, possibly biasing the steps involved in duplex unwinding and strand retention by RISC.	Cell	2003/10/17
Si2522	NM_001632	Khovorova	SEAP	250	UCGCUCUCGGUAACAUCCGgc	CGGAUGUUACCGAGAGCGA	-2.4	-2.4	-2.5	3	3	3	-40.8	53	-2.6	3	II	57.9	66.9	32	Flow cytometry and immunofluorescence	HEK293	0.25ug/Ml	24h	Functional siRNAs and miRNAs Exhibit Strand Biasunctional siRNAs and miRNAs Exhibit Strand Bias	14567918	2003	Both microRNAs (miRNA) and small interfering RNAs (siRNA) share a common set of cellular proteins (Dicer and the RNA-induced silencing complex [RISC]) to elicit RNA interference. In the following work, a statistical analysis of the internal stability of published miRNA sequences in the context of miRNA precursor hairpins revealed enhanced flexibility of miRNA precursors, especially at the 5′-anti-sense (AS) terminal base pair. The same trend was observed in siRNA, with functional duplexes displaying a lower internal stability (Δ0.5 kcal/mol) at the 5′-AS end than nonfunctional duplexes. Average internal stability of siRNA molecules retrieved from plant cells after introduction of long RNA sequences also shows this characteristic thermodynamic signature. Together, these results suggest that the thermodynamic properties of siRNA play a critical role in determining the molecule's function and longevity, possibly biasing the steps involved in duplex unwinding and strand retention by RISC.	Cell	2003/10/17
Si2523	M15077	Reynolds	Firefly luciferase	116160065	CUUGACUGGCGACGUAAUCca	GAUUACGUCGCCAGUCAAG	-2.4	-2.1	-0.8	0	0	4	-37.9	53	10.5	4	II	52.6	48.1	96	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2524	M15077	Reynolds	Firefly luciferase	116160065	UGCGUCGAGUUUUCCGGUAag	UACCGGAAAACUCGACGCA	-1.3	-2.1	-0.2	1	-1	4	-39.2	56.2	-0.7	5	II	52.6	46.6	95.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2525	M15077	Reynolds	Firefly luciferase	116160065	UCUGAUUUUUCUUGCGUCGag	CGACGCAAGAAAAAUCAGA	-2.4	-2.4	1	0	3	3	-34	80.5	3.4	7	Ia	42.1	68.9	95.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2526	M15077	Reynolds	Firefly luciferase	116160065	AACUCCUCCGCGCAACUUUuu	AAAGUUGCGCGGAGGAGUU	-0.9	-0.9	0.8	3	0	6	-39.4	64.7	-3.9	6	II	52.6	51.5	95.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2527	M15077	Reynolds	Firefly luciferase	116160065	UCGUCCACAAACACAACUCcu	GAGUUGUGUUUGUGGACGA	-2.4	-2.4	-0.4	0	2	2	-37.1	72.1	4.8	6	II	47.4	68	95.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2528	M15077	Reynolds	Firefly luciferase	116160065	UCUCUCUGAUUUUUCUUGCgu	GCAAGAAAAAUCAGAGAGA	-3.4	-2.4	1	2	4	2	-33.6	93	12.8	7	Ib	36.8	70.3	95.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2529	M15077	Reynolds	Firefly luciferase	116160065	UUUCCGGUAAGACCUUUCGgu	CGAAAGGUCUUACCGGAAA	-2.4	-0.9	-1.5	1	4	4	-36.3	78.8	2.4	7	II	47.4	73.9	95	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2530	M15077	Reynolds	Firefly luciferase	116160065	CUUGCGUCGAGUUUUCCGGua	CCGGAAAACUCGACGCAAG	-3.3	-2.1	-0.2	1	3	4	-38.7	69.5	3.4	3	II	57.9	57.7	94.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2531	M15077	Reynolds	Firefly luciferase	116160065	UCCGGUAAGACCUUUCGGUac	ACCGAAAGGUCUUACCGGA	-2.2	-2.4	-4.7	1	2	4	-40	60.3	-6.3	4	II	52.6	57.7	94.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2532	M15077	Reynolds	Firefly luciferase	116160065	UUUUCUUGCGUCGAGUUUUcc	AAAACUCGACGCAAGAAAA	-0.9	-0.9	0.9	2	1	3	-32.4	84.8	-6.3	8	II	36.8	69.1	94.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2533	M15077	Reynolds	Firefly luciferase	116160065	CGUCGAGUUUUCCGGUAAGac	CUUACCGGAAAACUCGACG	-2.1	-2.4	-0.2	-1	-1	4	-36.7	42.7	-2	2	II	52.6	32.8	94.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2534	M15077	Reynolds	Firefly luciferase	116160065	UGACUGGCGACGUAAUCCAcg	UGGAUUACGUCGCCAGUCA	-2.1	-2.1	-0.8	2	1	4	-40.3	53.2	-6.1	7	II	52.6	58.2	93.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2535	M15077	Reynolds	Firefly luciferase	116160065	ACAACUCCUCCGCGCAACUuu	AGUUGCGCGGAGGAGUUGU	-2.1	-2.2	0.8	0	1	6	-41.9	48.5	-3.9	3	II	57.9	46.7	93.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2536	M15077	Reynolds	Firefly luciferase	116160065	CCACAAACACAACUCCUCCgc	GGAGGAGUUGUGUUUGUGG	-3.3	-3.3	2.7	0	1	2	-38.8	53.3	15.1	4	II	52.6	52.2	93.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2537	M15077	Reynolds	Firefly luciferase	116160065	UCUUUUUCCGUCAUCGUCUuu	AGACGAUGACGGAAAAAGA	-2.1	-2.4	0.3	2	2	3	-35.5	85.2	-6.6	8	II	42.1	71.7	93.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2538	M15077	Reynolds	Firefly luciferase	116160065	AAGACCUUUCGGUACUUCGuc	CGAAGUACCGAAAGGUCUU	-2.4	-0.9	-4.9	2	3	3	-36.4	81.6	0	5	Ib	47.4	68.3	93.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2539	M15077	Reynolds	Firefly luciferase	116160065	UACUUGACUGGCGACGUAAuc	UUACGUCGCCAGUCAAGUA	-0.9	-1.3	1.3	0	-1	4	-37.9	64	-6.1	8	II	47.4	55.1	92.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2540	M15077	Reynolds	Firefly luciferase	116160065	UUCUUGCGUCGAGUUUUCCgg	GGAAAACUCGACGCAAGAA	-3.3	-0.9	0.9	0	4	3	-36.3	75.9	9.7	7	Ib	47.4	68.6	92.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2541	M15077	Reynolds	Firefly luciferase	116160065	CCUUUAUGAGGAUCUCUCUga	AGAGAGAUCCUCAUAAAGG	-2.1	-3.3	-2.3	0	-2	2	-36.5	70.9	-6	4	II	42.1	44.3	92.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2542	M15077	Reynolds	Firefly luciferase	116160065	ACAAACACAACUCCUCCGCgc	GCGGAGGAGUUGUGUUUGU	-3.4	-2.2	0.7	1	4	4	-39.2	64.1	13.2	7	Ia	52.6	62.8	92.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2543	M15077	Reynolds	Firefly luciferase	116160065	UAUGAGGAUCUCUCUGAUUuu	AAUCAGAGAGAUCCUCAUA	-0.9	-1.3	-2.3	0	1	2	-35.8	70.2	-3.6	6	II	36.8	63.7	91.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2544	M15077	Reynolds	Firefly luciferase	116160065	AUCCACGAUCUCUUUUUCCgu	GGAAAAAGAGAUCGUGGAU	-3.3	-1.1	1.2	2	4	2	-35.3	76.3	10.1	6	II	42.1	66.9	91.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2545	M15077	Reynolds	Firefly luciferase	116160065	UAAUCCACGAUCUCUUUUUcc	AAAAAGAGAUCGUGGAUUA	-0.9	-1.3	0.4	1	1	2	-31.8	93.1	-6.3	9	II	31.6	70.2	91.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2546	M15077	Reynolds	Firefly luciferase	116160065	UACUUCGUCCACAAACACAac	UGUGUUUGUGGACGAAGUA	-2.1	-1.3	-0.4	1	0	2	-36	72.9	-6.4	7	II	42.1	62.9	90.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2547	M15077	Reynolds	Firefly luciferase	116160065	AUUUUUCUUGCGUCGAGUUuu	AACUCGACGCAAGAAAAAU	-0.9	-1.1	0.9	1	2	3	-32.6	83.1	-6.3	6	II	36.8	57.1	90.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2548	M15077	Reynolds	Firefly luciferase	116160065	CUUCGUCCACAAACACAACuc	GUUGUGUUUGUGGACGAAG	-2.2	-2.1	-0.5	0	0	2	-35.6	55.5	5.1	2	II	47.4	46.6	90.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2549	M15077	Reynolds	Firefly luciferase	116160065	GAGUUUUCCGGUAAGACCUuu	AGGUCUUACCGGAAAACUC	-2.1	-2.4	-1.5	2	0	4	-37.3	71.5	-3.5	3	II	47.4	54	90.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2550	M15077	Reynolds	Firefly luciferase	116160065	GACCUUUCGGUACUUCGUCca	GACGAAGUACCGAAAGGUC	-2.4	-2.4	-4.9	3	2	3	-38	61.1	12.1	2	II	52.6	50.8	89.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2551	M15077	Reynolds	Firefly luciferase	116160065	UGCUCCAAAACAACAACGGcg	CCGUUGUUGUUUUGGAGCA	-3.3	-2.1	-0.5	0	3	3	-36.4	67	-4.9	6	II	47.4	65.8	89.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2552	M15077	Reynolds	Firefly luciferase	116160065	UUUCCGUGCUCCAAAACAAca	UUGUUUUGGAGCACGGAAA	-0.9	-0.9	-1.4	3	1	3	-35.4	63.3	-11	5	II	42.1	57.5	89.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2553	M15077	Reynolds	Firefly luciferase	116160065	ACGAUCUCUUUUUCCGUCAuc	UGACGGAAAAAGAGAUCGU	-2.1	-2.2	0.6	2	-1	3	-35.5	76	-0.7	6	II	42.1	53.7	88.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2554	M15077	Reynolds	Firefly luciferase	116160065	UCGCGGUUGUUACUUGACUgg	AGUCAAGUAACAACCGCGA	-2.1	-2.4	1.1	3	1	5	-37.5	59.6	0.8	4	II	47.4	57.9	88.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2555	M15077	Reynolds	Firefly luciferase	116160065	UUUCGCGGUUGUUACUUGAcu	UCAAGUAACAACCGCGAAA	-2.4	-0.9	1.2	1	1	5	-35	73.7	-0.7	7	II	42.1	58.3	88.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2556	M15077	Reynolds	Firefly luciferase	116160065	UGAGGAUCUCUCUGAUUUUuc	AAAAUCAGAGAGAUCCUCA	-0.9	-2.1	-2.3	0	0	2	-35.2	64.8	-6.3	5	II	36.8	55.8	88.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2557	M15077	Reynolds	Firefly luciferase	116160065	UUUCCGUCAUCGUCUUUCCgu	GGAAAGACGAUGACGGAAA	-3.3	-0.9	0.3	2	4	3	-36.7	83.5	5.1	7	Ib	47.4	73.7	87.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2558	M15077	Reynolds	Firefly luciferase	116160065	UCUCUGAUUUUUCUUGCGUcg	ACGCAAGAAAAAUCAGAGA	-2.2	-2.4	1	1	3	3	-33.7	79.2	3.8	6	II	36.8	61.5	87.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2559	M15077	Reynolds	Firefly luciferase	116160065	GUCCACAAACACAACUCCUcc	AGGAGUUGUGUUUGUGGAC	-2.1	-2.2	1.5	1	0	2	-37.7	53.4	-3.9	3	III	47.4	46.1	87.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2560	M15077	Reynolds	Firefly luciferase	116160065	ACUGGCGACGUAAUCCACGau	CGUGGAUUACGUCGCCAGU	-2.4	-2.2	-1.4	2	4	4	-40.4	45.9	-0.3	2	II	57.9	60.9	87.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2561	M15077	Reynolds	Firefly luciferase	116160065	AAACACAACUCCUCCGCGCaa	GCGCGGAGGAGUUGUGUUU	-3.4	-0.9	-2.6	3	5	6	-40.7	56	5.4	4	Ia	57.9	64.3	86.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2562	M15077	Reynolds	Firefly luciferase	116160065	AGGAUCUCUCUGAUUUUUCuu	GAAAAAUCAGAGAGAUCCU	-2.4	-2.1	0.7	2	2	2	-34	82.7	7.1	6	Ib	36.8	63.6	86.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2563	M15077	Reynolds	Firefly luciferase	116160065	ACUUUUUCGCGGUUGUUACuu	GUAACAACCGCGAAAAAGU	-2.2	-2.2	0.9	3	3	5	-33.6	88.8	9.7	7	Ia	42.1	57.8	85.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2564	M15077	Reynolds	Firefly luciferase	116160065	GUUACUUGACUGGCGACGUaa	ACGUCGCCAGUCAAGUAAC	-2.2	-2.2	1.9	1	1	4	-38.8	58.8	-4	4	II	52.6	46.3	84.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2565	M15077	Reynolds	Firefly luciferase	116160065	GACGUAAUCCACGAUCUCUuu	AGAGAUCGUGGAUUACGUC	-2.1	-2.4	-0.1	1	-1	2	-37.4	58.2	-4	4	II	47.4	45.2	82.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2566	M15077	Reynolds	Firefly luciferase	116160065	UCCGUGCUCCAAAACAACAac	UGUUGUUUUGGAGCACGGA	-2.1	-2.4	-0.3	1	-1	3	-37.9	55.6	-4	4	II	47.4	55.2	82.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2567	M15077	Reynolds	Firefly luciferase	116160065	UCUCUUUUUCCGUCAUCGUcu	ACGAUGACGGAAAAAGAGA	-2.2	-2.4	0.3	2	2	3	-35.5	78	-6.3	6	II	42.1	60.9	80.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2568	M15077	Reynolds	Firefly luciferase	116160065	CGGUAAGACCUUUCGGUACuu	GUACCGAAAGGUCUUACCG	-2.2	-2.4	-4.9	0	-2	3	-37.8	57.5	11.1	3	II	52.6	41.3	80.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2569	M15077	Reynolds	Firefly luciferase	116160065	GGUUGUUACUUGACUGGCGac	CGCCAGUCAAGUAACAACC	-2.4	-3.3	1.1	2	2	4	-38.1	58.7	-2.6	3	II	52.6	48.1	79.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2570	M15077	Reynolds	Firefly luciferase	116160065	CCUCCGCGCAACUUUUUCGcg	CGAAAAAGUUGCGCGGAGG	-2.4	-3.3	1.5	1	1	6	-38.4	62.9	0.3	2	II	57.9	47.2	78.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2571	M15077	Reynolds	Firefly luciferase	116160065	UUGGCCUUUAUGAGGAUCUcu	AGAUCCUCAUAAAGGCCAA	-2.1	-0.9	-0.5	0	1	4	-37.2	75.1	-6.6	7	II	42.1	71	78.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2572	M15077	Reynolds	Firefly luciferase	116160065	UUUCGGUACUUCGUCCACAaa	UGUGGACGAAGUACCGAAA	-2.1	-0.9	-0.7	3	2	3	-37.7	62.1	-3.8	6	II	47.4	64.9	77.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2573	M15077	Reynolds	Firefly luciferase	116160065	UUUUUCGCGGUUGUUACUUga	AAGUAACAACCGCGAAAAA	-0.9	-0.9	1.2	1	3	5	-32.3	83.6	1.4	8	II	36.8	67	76.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2574	M15077	Reynolds	Firefly luciferase	116160065	CUCCAAAACAACAACGGCGgc	CGCCGUUGUUGUUUUGGAG	-2.4	-2.1	-0.7	1	1	5	-36.7	51.5	-1.9	4	II	52.6	57.9	74.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2575	M15077	Reynolds	Firefly luciferase	116160065	UUUUUCCGUCAUCGUCUUUcc	AAAGACGAUGACGGAAAAA	-0.9	-0.9	0.3	-1	1	3	-32.8	84.4	3.8	8	II	36.8	57.1	73.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2576	M15077	Reynolds	Firefly luciferase	116160065	UCCGUCAUCGUCUUUCCGUgc	ACGGAAAGACGAUGACGGA	-2.2	-2.4	-1	1	2	3	-39.5	66.9	-4	4	II	52.6	60.2	73.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2577	M15077	Reynolds	Firefly luciferase	116160065	GUUUUCCGGUAAGACCUUUcg	AAAGGUCUUACCGGAAAAC	-0.9	-2.2	-1.5	1	-1	4	-34.6	61	-1.7	4	II	42.1	38.7	71.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2578	M15077	Reynolds	Firefly luciferase	116160065	CCUUUCGGUACUUCGUCCAca	UGGACGAAGUACCGAAAGG	-2.1	-3.3	-3.4	-1	-2	3	-38.8	63.7	-5.1	4	III	52.6	40.3	70.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2579	M15077	Reynolds	Firefly luciferase	116160065	GAUCUCUCUGAUUUUUCUUgc	AAGAAAAAUCAGAGAGAUC	-0.9	-2.4	0.7	2	1	1	-31.6	90.6	3.7	4	II	31.6	48.7	68.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2580	M15077	Reynolds	Firefly luciferase	116160065	GCGACGUAAUCCACGAUCUcu	AGAUCGUGGAUUACGUCGC	-2.1	-3.4	-0.6	1	-2	3	-38.7	52.3	-11.3	3	III	52.6	39.9	67	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2581	M15077	Reynolds	Firefly luciferase	116160065	GCGGUUGUUACUUGACUGGcg	CCAGUCAAGUAACAACCGC	-3.3	-3.4	1.1	-1	0	4	-38.1	56.9	0.7	4	II	52.6	44.4	66.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2582	M15077	Reynolds	Firefly luciferase	116160065	GUAAGACCUUUCGGUACUUcg	AAGUACCGAAAGGUCUUAC	-0.9	-2.2	-4.9	-1	0	3	-35.1	46.1	-6.3	2	II	42.1	36.3	64.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2583	M15077	Reynolds	Firefly luciferase	116160065	AACGGCGGCGGGAAGUUCAcc	UGAACUUCCCGCCGCCGUU	-2.1	-0.9	1.6	2	0	10	-43.3	41.3	-6.3	4	II	63.2	47.5	63.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2584	M15077	Reynolds	Firefly luciferase	116160065	UCGAGUUUUCCGGUAAGACcu	GUCUUACCGGAAAACUCGA	-2.2	-2.4	-0.2	1	3	4	-36.7	67	7.4	5	Ib	47.4	62.1	61	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2585	M15077	Reynolds	Firefly luciferase	116160065	CCACGAUCUCUUUUUCCGUca	ACGGAAAAAGAGAUCGUGG	-2.2	-3.3	0.3	1	0	3	-36.4	57.8	-0.6	2	III	47.4	42.8	58.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2586	M15077	Reynolds	Firefly luciferase	116160065	AUCGUCUUUCCGUGCUCCAaa	UGGAGCACGGAAAGACGAU	-2.1	-1.1	-1	1	1	3	-40.1	61.1	-6.1	4	II	52.6	51.3	52.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2587	M15077	Reynolds	Firefly luciferase	116160065	GAUCUCUUUUUCCGUCAUCgu	GAUGACGGAAAAAGAGAUC	-2.4	-2.4	2.6	1	1	3	-34.4	69	7.5	4	II	42.1	49	47.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2588	M15077	Reynolds	Firefly luciferase	116160065	CGUAAUCCACGAUCUCUUUuu	AAAGAGAUCGUGGAUUACG	-0.9	-2.4	0.4	-1	-3	2	-34.6	67	-3.3	4	II	42.1	37.6	47.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2589	M15077	Reynolds	Firefly luciferase	116160065	CAACAACGGCGGCGGGAAGuu	CUUCCCGCCGCCGUUGUUG	-2.1	-2.1	1.6	0	0	10	-43	26.4	3.1	2	II	68.4	30.6	47.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2590	M15077	Reynolds	Firefly luciferase	116160065	UUGUUACUUGACUGGCGACgu	GUCGCCAGUCAAGUAACAA	-2.2	-0.9	1.1	-1	3	4	-37.2	71.3	7.4	6	Ia	47.4	60.7	44.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2591	M15077	Reynolds	Firefly luciferase	116160065	UGGCGACGUAAUCCACGAUcu	AUCGUGGAUUACGUCGCCA	-1.1	-2.1	-1.4	1	0	4	-39.6	51	3.8	3	II	52.6	43.8	44.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2592	M15077	Reynolds	Firefly luciferase	116160065	AACAACAACGGCGGCGGGAag	UCCCGCCGCCGUUGUUGUU	-2.4	-0.9	1.6	1	1	10	-43.1	37.8	-3.4	5	II	63.2	43	43.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2593	M15077	Reynolds	Firefly luciferase	116160065	CGUCAUCGUCUUUCCGUGCuc	GCACGGAAAGACGAUGACG	-3.4	-2.4	-1	0	0	3	-39.3	56.5	11.1	1	II	57.9	46.1	39.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2594	M15077	Reynolds	Firefly luciferase	116160065	CUCCUCCGCGCAACUUUUUcg	AAAAAGUUGCGCGGAGGAG	-0.9	-2.1	1.5	1	-3	6	-38.1	51.7	-10.9	3	III	52.6	37.8	38.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2595	M15077	Reynolds	Firefly luciferase	116160065	GGCCUUUAUGAGGAUCUCUcu	AGAGAUCCUCAUAAAGGCC	-2.1	-3.3	-0.7	2	-2	4	-38.7	58.1	-1.7	3	III	47.4	41.3	38.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2596	M15077	Reynolds	Firefly luciferase	116160065	GGUACUUCGUCCACAAACAca	UGUUUGUGGACGAAGUACC	-2.1	-3.3	-0.4	2	-2	2	-37.2	59.2	-11	3	II	47.4	39.3	35.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2597	M15077	Reynolds	Firefly luciferase	116160065	CGUGCUCCAAAACAACAACgg	GUUGUUGUUUUGGAGCACG	-2.2	-2.4	-0.3	-1	-1	2	-35.3	59.1	10.1	2	II	47.4	43.5	32.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2598	M15077	Reynolds	Firefly luciferase	116160065	UCCGCGCAACUUUUUCGCGgu	CGCGAAAAAGUUGCGCGGA	-2.4	-2.4	-4.4	0	4	6	-38.8	60.2	5.4	1	II	57.9	58.4	32.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2599	M15077	Reynolds	Firefly luciferase	116160065	CGCGCAACUUUUUCGCGGUug	ACCGCGAAAAAGUUGCGCG	-2.2	-2.4	-3.9	-1	-2	5	-38.6	47.9	-3	-1	III	57.9	34.3	29.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2600	M15077	Reynolds	Firefly luciferase	116160065	ACACAACUCCUCCGCGCAAcu	UUGCGCGGAGGAGUUGUGU	-0.9	-2.2	-2.7	2	0	6	-41.9	32	-1	2	II	57.9	31.8	24.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2601	M15077	Reynolds	Firefly luciferase	116160065	UCAUCGUCUUUCCGUGCUCca	GAGCACGGAAAGACGAUGA	-2.4	-2.4	-1	0	3	3	-39.2	61.5	7.5	5	Ib	52.6	61.7	20.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2602	M15077	Reynolds	Firefly luciferase	116160065	ACAACGGCGGCGGGAAGUUca	AACUUCCCGCCGCCGUUGU	-0.9	-2.2	1.6	1	1	10	-43.1	27.6	-6.3	4	II	63.2	39	18.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2603	M15077	Reynolds	Firefly luciferase	116160065	CCAAAACAACAACGGCGGCgg	GCCGCCGUUGUUGUUUUGG	-3.4	-3.3	0.4	-2	1	7	-38.9	35.9	12.8	2	II	57.9	37.5	13.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2604	M15077	Reynolds	Firefly luciferase	116160065	CGCAACUUUUUCGCGGUUGuu	CAACCGCGAAAAAGUUGCG	-2.1	-2.4	-3.5	0	-2	5	-35.8	56.9	-1.7	3	II	52.6	44	-1.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2605	M15077	Reynolds	Firefly luciferase	116160065	CAACUUUUUCGCGGUUGUUac	AACAACCGCGAAAAAGUUG	-0.9	-2.1	-0.3	0	-1	5	-33.1	62.8	-8.3	4	II	42.1	43.3	-5.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2606	M15077	Reynolds	Firefly luciferase	116160065	CGUCUUUCCGUGCUCCAAAac	UUUGGAGCACGGAAAGACG	-0.9	-2.4	-0.3	2	-3	3	-38.4	57.9	-1	3	II	52.6	37.2	-10.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2607	M15077	Reynolds	Firefly luciferase	116160065	UCUUGGCCUUUAUGAGGAUcu	AUCCUCAUAAAGGCCAAGA	-1.1	-2.4	-0.5	-1	1	4	-37.2	55.1	-4	5	II	42.1	42.9	-12.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2608	M15077	Reynolds	Firefly luciferase	116160065	UCGGUACUUCGUCCACAAAca	UUUGUGGACGAAGUACCGA	-0.9	-2.4	-0.7	0	-2	3	-37.7	57.4	1.7	4	II	47.4	40.6	-18	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2609	M15077	Reynolds	Firefly luciferase	116160065	UCUUUCCGUGCUCCAAAACaa	GUUUUGGAGCACGGAAAGA	-2.2	-2.4	-1.4	0	2	3	-36.9	72.4	10.5	5	Ib	47.4	57	-27.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2610	M60857	Reynolds	Cyclophilin B	829639	CUCUCCUGUAGCUAAGGCCac	GGCCUUAGCUACAGGAGAG	-3.3	-2.1	-1.1	0	1	4	-41.9	66	3.1	2	II	57.9	56.4	94.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2611	M60857	Reynolds	Cyclophilin B	829639	UUUGUAGCCAAAUCCUUUCuc	GAAAGGAUUUGGCUACAAA	-2.4	-0.9	-0.5	1	3	3	-33.2	88.5	9.7	9	Ia	36.8	78.2	94.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2612	M60857	Reynolds	Cyclophilin B	829639	AAUCCUUUCUCUCCUGUAGcu	CUACAGGAGAGAAAGGAUU	-2.1	-0.9	1.8	4	3	2	-36.1	86.2	3.1	6	Ia	42.1	59.6	93.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2613	M60857	Reynolds	Cyclophilin B	829639	UAGCCAAAUCCUUUCUCUCcu	GAGAGAAAGGAUUUGGCUA	-2.4	-1.3	1.5	1	4	3	-36.1	80.1	10.4	5	Ib	42.1	70.6	93.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2614	M60857	Reynolds	Cyclophilin B	829639	CACCGUAGAUGCUCUUUCCuc	GGAAAGAGCAUCUACGGUG	-3.3	-2.1	0.2	0	0	3	-38.7	64.9	3.1	3	II	52.6	52.1	93.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2615	M60857	Reynolds	Cyclophilin B	829639	UCUCCGCCCUGGAUCAUGAag	UCAUGAUCCAGGGCGGAGA	-2.4	-2.4	-2.2	2	1	6	-43.7	55.2	-4	3	II	57.9	48	92.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2616	M60857	Reynolds	Cyclophilin B	829639	GAAGUCUCCGCCCUGGAUCau	GAUCCAGGGCGGAGACUUC	-2.4	-2.4	-2.2	2	2	6	-43.6	56.3	7.4	2	II	63.2	52.2	92.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2617	M60857	Reynolds	Cyclophilin B	829639	GUUCUCAUCGGGGAAGCGCuc	GCGCUUCCCCGAUGAGAAC	-3.4	-2.2	1	2	3	5	-42.5	48.1	7.4	3	II	63.2	50.5	92.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2618	M60857	Reynolds	Cyclophilin B	829639	UGUAGCCAAAUCCUUUCUCuc	GAGAAAGGAUUUGGCUACA	-2.4	-2.1	1.4	0	4	3	-35.9	77.6	9.8	6	II	42.1	65.4	92.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2619	M60857	Reynolds	Cyclophilin B	829639	GAUUACACGAUGGAAUUUGcu	CAAAUUCCAUCGUGUAAUC	-2.1	-2.4	2.9	1	1	2	-31.8	69	2.7	7	II	36.8	53.3	92.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2620	M60857	Reynolds	Cyclophilin B	829639	ACACGAUGGAAUUUGCUGUuu	ACAGCAAAUUCCAUCGUGU	-2.2	-2.2	1.8	4	2	2	-35.9	70.4	3.7	6	II	42.1	54.4	91.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2621	M60857	Reynolds	Cyclophilin B	829639	AUGCUCUUUCCUCCUGUGCca	GCACAGGAGGAAAGAGCAU	-3.4	-1.1	0.2	3	3	2	-40.4	75.1	10.5	6	Ib	52.6	64.3	90.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2622	M60857	Reynolds	Cyclophilin B	829639	AAAUUAUCCACUGUUUUUGga	CAAAAACAGUGGAUAAUUU	-2.1	-0.9	2.2	3	4	2	-28.3	98.5	3	9	Ia	26.3	71.1	90.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2623	M60857	Reynolds	Cyclophilin B	829639	UCAGUUUGAAGUUCUCAUCgg	GAUGAGAACUUCAAACUGA	-2.4	-2.4	0.4	1	2	1	-33.7	89.4	10.4	7	Ia	36.8	73.5	89.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2624	M60857	Reynolds	Cyclophilin B	829639	CCUGGUGAAGUCUCCGCCCug	GGGCGGAGACUUCACCAGG	-3.3	-3.3	-0.8	-1	1	6	-45.3	39.9	3.1	1	II	68.4	50.5	89	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2625	M60857	Reynolds	Cyclophilin B	829639	GAAGCGCUCACCGUAGAUGcu	CAUCUACGGUGAGCGCUUC	-2.1	-2.4	-0.7	1	2	4	-40.3	48.6	-2.4	2	II	57.9	46.2	89	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2626	M60857	Reynolds	Cyclophilin B	829639	UGAAGUCCUUGAUUACACGau	CGUGUAAUCAAGGACUUCA	-2.4	-2.1	1.5	0	4	2	-35.1	74.4	2.3	5	Ib	42.1	67.8	88.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2627	M60857	Reynolds	Cyclophilin B	829639	UCCGCCCUGGAUCAUGAAGuc	CUUCAUGAUCCAGGGCGGA	-2.1	-2.4	-1.4	1	1	6	-42.2	58.8	7.7	1	II	57.9	49.3	88.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2628	M60857	Reynolds	Cyclophilin B	829639	GCCAAAUCCUUUCUCUCCUgu	AGGAGAGAAAGGAUUUGGC	-2.1	-3.4	2.3	2	0	3	-38.1	62.7	6.5	3	II	47.4	40.6	88.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2629	M60857	Reynolds	Cyclophilin B	829639	UAUCCACUGUUUUUGGAACag	GUUCCAAAAACAGUGGAUA	-2.2	-1.3	-2.5	2	4	2	-33.3	85.9	12.8	4	Ib	36.8	62.7	88.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2630	M60857	Reynolds	Cyclophilin B	829639	AUCAUGAAGUCCUUGAUUAca	UAAUCAAGGACUUCAUGAU	-1.3	-1.1	0.7	2	0	2	-32.9	79.7	-6.1	7	II	31.6	55.5	87.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2631	M60857	Reynolds	Cyclophilin B	829639	AUCUCCCCUGGUGAAGUCUcc	AGACUUCACCAGGGGAGAU	-2.1	-1.1	-2.5	1	1	4	-41.8	63	-1	4	II	52.6	50.3	87.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2632	M60857	Reynolds	Cyclophilin B	829639	UUGAUUACACGAUGGAAUUug	AAUUCCAUCGUGUAAUCAA	-0.9	-0.9	1.5	-1	0	2	-31.8	83	-4	7	II	31.6	65.9	87.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2633	M60857	Reynolds	Cyclophilin B	829639	GUCCUUGAUUACACGAUGGaa	CCAUCGUGUAAUCAAGGAC	-3.3	-2.2	0.9	1	1	2	-36.5	63.6	-2.6	4	II	47.4	51.6	87.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2634	M60857	Reynolds	Cyclophilin B	829639	AAGUCCUUGAUUACACGAUgg	AUCGUGUAAUCAAGGACUU	-1.1	-0.9	1.5	3	1	2	-34.1	78.1	-3.5	6	II	36.8	55.8	86.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2635	M60857	Reynolds	Cyclophilin B	829639	UCUCAUCGGGGAAGCGCUCac	GAGCGCUUCCCCGAUGAGA	-2.4	-2.4	1	1	3	5	-43.9	39.5	7.5	4	Ib	63.2	54.7	86.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2636	M60857	Reynolds	Cyclophilin B	829639	UUACACGAUGGAAUUUGCUgu	AGCAAAUUCCAUCGUGUAA	-2.1	-0.9	1.1	1	4	2	-33.8	71.2	-1.5	7	II	36.8	68.7	86.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2637	M60857	Reynolds	Cyclophilin B	829639	AGUUUGAAGUUCUCAUCGGgg	CCGAUGAGAACUUCAAACU	-3.3	-2.1	0.1	2	4	3	-34.9	69.4	-2.4	6	Ia	42.1	59.5	85.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2638	M60857	Reynolds	Cyclophilin B	829639	CAAAUCCUUUCUCUCCUGUag	ACAGGAGAGAAAGGAUUUG	-2.2	-2.1	2.3	-1	0	2	-35.7	75.9	-0.6	5	II	42.1	47.9	84.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2639	M60857	Reynolds	Cyclophilin B	829639	UCUUUCCUCCUGUGCCAUCuc	GAUGGCACAGGAGGAAAGA	-2.4	-2.4	3	0	2	3	-40.6	67.9	7.4	6	Ib	52.6	59.1	84.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2640	M60857	Reynolds	Cyclophilin B	829639	UUUGAAGUUCUCAUCGGGGaa	CCCCGAUGAGAACUUCAAA	-3.3	-0.9	0.1	0	6	5	-37.2	65.1	0.1	7	Ia	47.4	72.9	84.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2641	M60857	Reynolds	Cyclophilin B	829639	UAAGGCCACAAAAUUAUCCac	GGAUAAUUUUGUGGCCUUA	-3.3	-1.3	0.9	1	5	4	-33.5	81.8	9.4	7	II	36.8	80.6	83.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2642	M60857	Reynolds	Cyclophilin B	829639	UCAUCGGGGAAGCGCUCACcg	GUGAGCGCUUCCCCGAUGA	-2.2	-2.4	0.7	0	3	5	-43.7	49.5	12.1	6	II	63.2	55.7	83.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2643	M60857	Reynolds	Cyclophilin B	829639	UGAAGUUCUCAUCGGGGAAgc	UUCCCCGAUGAGAACUUCA	-0.9	-2.1	0.1	0	0	5	-38.7	63.9	4	6	II	47.4	48.4	83.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2644	M60857	Reynolds	Cyclophilin B	829639	UGUUUUUGUAGCCAAAUCCuu	GGAUUUGGCUACAAAAACA	-3.3	-2.1	-1.3	1	3	3	-33	93.5	7.5	8	Ia	36.8	79.5	83.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2645	M60857	Reynolds	Cyclophilin B	829639	CCCUGGAUCAUGAAGUCCUug	AGGACUUCAUGAUCCAGGG	-2.1	-3.3	-2.1	0	-2	3	-40.7	49.8	-8.6	2	III	52.6	46	82.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2646	M60857	Reynolds	Cyclophilin B	829639	UUUUUGUAGCCAAAUCCUUuc	AAGGAUUUGGCUACAAAAA	-0.9	-0.9	-1.3	2	2	3	-31.7	78.3	-8.9	7	II	31.6	69.2	81.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2647	M60857	Reynolds	Cyclophilin B	829639	UAGCUAAGGCCACAAAAUUau	AAUUUUGUGGCCUUAGCUA	-0.9	-1.3	-1.6	2	0	4	-34.3	61.5	-3.9	5	II	36.8	60.3	81.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2648	M60857	Reynolds	Cyclophilin B	829639	CUCCCCUGGUGAAGUCUCCgc	GGAGACUUCACCAGGGGAG	-3.3	-2.1	-1.6	1	0	4	-44	46.8	2.8	1	II	63.2	46.7	80.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2649	M60857	Reynolds	Cyclophilin B	829639	UCCUCCUGUGCCAUCUCCCcu	GGGAGAUGGCACAGGAGGA	-3.3	-2.4	3	1	4	3	-45.4	64.1	2.4	4	II	63.2	66.8	80.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2650	M60857	Reynolds	Cyclophilin B	829639	AUCGGGGAAGCGCUCACCGua	CGGUGAGCGCUUCCCCGAU	-2.4	-1.1	-1.2	1	5	5	-44.9	41.1	2.4	2	II	68.4	60.6	80.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2651	M60857	Reynolds	Cyclophilin B	829639	UGGAAUUUGCUGUUUUUGUag	ACAAAAACAGCAAAUUCCA	-2.2	-2.1	4.5	2	1	2	-31.4	91.1	1.1	7	II	31.6	65.2	78.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2652	M60857	Reynolds	Cyclophilin B	829639	AUUUGCUGUUUUUGUAGCCaa	GGCUACAAAAACAGCAAAU	-3.3	-1.1	-1.9	1	5	3	-32.8	83.4	10.4	5	Ia	36.8	67.3	77.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2653	M60857	Reynolds	Cyclophilin B	829639	UUCUCUCCUGUAGCUAAGGcc	CCUUAGCUACAGGAGAGAA	-3.3	-0.9	-0.7	0	3	2	-38.5	70.7	0	5	Ib	47.4	67.9	77.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2654	M60857	Reynolds	Cyclophilin B	829639	CCCCUGGUGAAGUCUCCGCcc	GCGGAGACUUCACCAGGGG	-3.4	-3.3	-0.8	1	0	4	-45.3	44.8	7.7	2	II	68.4	44.7	76.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2655	M60857	Reynolds	Cyclophilin B	829639	CUCCUGUGCCAUCUCCCCUgg	AGGGGAGAUGGCACAGGAG	-2.1	-2.1	3	2	0	4	-45.1	56.2	4.1	2	III	63.2	48.7	76.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2656	M60857	Reynolds	Cyclophilin B	829639	UCCUUUCUCUCCUGUAGCUaa	AGCUACAGGAGAGAAAGGA	-2.1	-2.4	1.6	0	1	2	-39.6	70.6	-8.7	5	II	47.4	52.1	75.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2657	M60857	Reynolds	Cyclophilin B	829639	AUUAUCCACUGUUUUUGGAac	UCCAAAAACAGUGGAUAAU	-2.4	-1.1	-2.3	2	3	2	-32.2	90.5	1.6	5	II	31.6	51.9	75.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2658	M60857	Reynolds	Cyclophilin B	829639	AGAUGCUCUUUCCUCCUGUgc	ACAGGAGGAAAGAGCAUCU	-2.2	-2.1	1.1	2	2	2	-39.4	67.5	-1.6	6	II	47.4	54.4	74.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2659	M60857	Reynolds	Cyclophilin B	829639	CUGGAUCAUGAAGUCCUUGau	CAAGGACUUCAUGAUCCAG	-2.1	-2.1	-2.1	-1	0	2	-37.1	56.1	5.3	2	II	47.4	52	74.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2660	M60857	Reynolds	Cyclophilin B	829639	UGUGCCAUCUCCCCUGGUGaa	CACCAGGGGAGAUGGCACA	-2.1	-2.1	-1.5	1	3	4	-44.9	58.8	-2.3	3	II	63.2	57.3	73.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2661	M60857	Reynolds	Cyclophilin B	829639	GUAGUGCUUCAGUUUGAAGuu	CUUCAAACUGAAGCACUAC	-2.1	-2.2	-2	0	2	2	-34.4	70.1	4.6	2	II	42.1	46.7	73.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2662	M60857	Reynolds	Cyclophilin B	829639	CGCUCACCGUAGAUGCUCUuu	AGAGCAUCUACGGUGAGCG	-2.1	-2.4	-0.9	0	-2	3	-41.5	51.7	-3.9	1	III	57.9	32.3	71.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2663	M60857	Reynolds	Cyclophilin B	829639	GAAUUUGCUGUUUUUGUAGcc	CUACAAAAACAGCAAAUUC	-2.1	-2.4	2.8	0	2	2	-29.4	90.1	5.4	6	II	31.6	55.7	71.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2664	M60857	Reynolds	Cyclophilin B	829639	UUUCCUCCUGUGCCAUCUCcc	GAGAUGGCACAGGAGGAAA	-2.4	-0.9	3	1	4	3	-40.6	67.4	9.8	5	Ib	52.6	67.3	70.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2665	M60857	Reynolds	Cyclophilin B	829639	AGUGCUUCAGUUUGAAGUUcu	AACUUCAAACUGAAGCACU	-0.9	-2.1	-2.5	1	1	2	-34	72	-1	3	II	36.8	50.5	69.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2666	M60857	Reynolds	Cyclophilin B	829639	UUGCUGUUUUUGUAGCCAAau	UUGGCUACAAAAACAGCAA	-0.9	-0.9	-1.9	2	0	3	-33.8	73.1	-3.8	5	II	36.8	60.1	68	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2667	M60857	Reynolds	Cyclophilin B	829639	AAGUUCUCAUCGGGGAAGCgc	GCUUCCCCGAUGAGAACUU	-3.4	-0.9	1.2	1	3	5	-39.7	61.7	5.1	6	Ia	52.6	57.5	67.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2668	M60857	Reynolds	Cyclophilin B	829639	GGAUCAUGAAGUCCUUGAUua	AUCAAGGACUUCAUGAUCC	-1.1	-3.3	-0.4	1	-1	2	-36.4	69.8	1.5	4	II	42.1	32.5	66.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2669	M60857	Reynolds	Cyclophilin B	829639	AGCGCUCACCGUAGAUGCUcu	AGCAUCUACGGUGAGCGCU	-2.1	-2.1	-0.9	1	1	4	-42.5	46.7	-3.5	1	II	57.9	39.4	64.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2670	M60857	Reynolds	Cyclophilin B	829639	CAUGAAGUCCUUGAUUACAcg	UGUAAUCAAGGACUUCAUG	-2.1	-2.1	1.6	0	-2	2	-33.7	61	0	4	II	36.8	45.2	63.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2671	M60857	Reynolds	Cyclophilin B	829639	CUCCUGUAGCUAAGGCCACaa	GUGGCCUUAGCUACAGGAG	-2.2	-2.1	-5.8	1	0	4	-41.7	39.8	2.8	1	II	57.9	45.8	60.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2672	M60857	Reynolds	Cyclophilin B	829639	GUAGAUGCUCUUUCCUCCUgu	AGGAGGAAAGAGCAUCUAC	-2.1	-2.2	0.4	0	1	2	-38.6	62.1	1.8	4	II	47.4	52.5	58.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2673	M60857	Reynolds	Cyclophilin B	829639	CCGUAGAUGCUCUUUCCUCcu	GAGGAAAGAGCAUCUACGG	-2.4	-3.3	0.6	0	0	3	-38.9	57	8.1	3	II	52.6	48.8	58.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2674	M60857	Reynolds	Cyclophilin B	829639	GCUCUUUCCUCCUGUGCCAuc	UGGCACAGGAGGAAAGAGC	-2.1	-3.4	1.9	2	-1	3	-42.6	50.4	-8.5	2	II	57.9	32.9	55.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2675	M60857	Reynolds	Cyclophilin B	829639	AGGCCACAAAAUUAUCCACug	GUGGAUAAUUUUGUGGCCU	-2.2	-2.1	0.9	2	2	4	-35.6	68.3	12.7	3	II	42.1	50.6	55	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2676	M60857	Reynolds	Cyclophilin B	829639	UGUAGCUAAGGCCACAAAAuu	UUUUGUGGCCUUAGCUACA	-0.9	-2.1	-5.9	1	-1	4	-36.6	68	-3.7	5	II	42.1	52.5	52	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2677	M60857	Reynolds	Cyclophilin B	829639	CGCCCUGGAUCAUGAAGUCcu	GACUUCAUGAUCCAGGGCG	-2.4	-2.4	1	-1	-1	5	-41.1	40.7	3.1	0	II	57.9	34.6	51.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2678	M60857	Reynolds	Cyclophilin B	829639	CUCACCGUAGAUGCUCUUUcc	AAAGAGCAUCUACGGUGAG	-0.9	-2.1	-0.9	-1	-3	3	-37.5	67.2	-0.7	4	III	47.4	42	50.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2679	M60857	Reynolds	Cyclophilin B	829639	CAAAAUUAUCCACUGUUUUug	AAAACAGUGGAUAAUUUUG	-0.9	-2.1	2.1	0	-1	2	-28.3	83.2	-0.9	6	II	26.3	54.4	49.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2680	M60857	Reynolds	Cyclophilin B	829639	AGUCUCCGCCCUGGAUCAUga	AUGAUCCAGGGCGGAGACU	-1.1	-2.1	-2.2	2	1	6	-43.5	42.2	-3.3	2	II	57.9	31.5	48.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2681	M60857	Reynolds	Cyclophilin B	829639	UGGUGAAGUCUCCGCCCUGga	CAGGGCGGAGACUUCACCA	-2.1	-2.1	-0.9	-1	2	6	-44.1	41.2	0.1	4	Ib	63.2	55.7	46.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2682	M60857	Reynolds	Cyclophilin B	829639	CCUUGAUUACACGAUGGAAuu	UUCCAUCGUGUAAUCAAGG	-0.9	-3.3	1.5	0	-3	2	-35.2	55.3	-5.8	3	II	42.1	28.4	45.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2683	M60857	Reynolds	Cyclophilin B	829639	UGCUUCAGUUUGAAGUUCUca	AGAACUUCAAACUGAAGCA	-2.1	-2.1	-2.5	0	0	2	-34.2	76.1	-6.6	6	II	36.8	61.7	45.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2684	M60857	Reynolds	Cyclophilin B	829639	CCAUCUCCCCUGGUGAAGUcu	ACUUCACCAGGGGAGAUGG	-2.2	-3.3	-2.5	0	-1	4	-42.7	50	-3.6	0	III	57.9	37.7	41.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2685	M60857	Reynolds	Cyclophilin B	829639	GUGAAGUCUCCGCCCUGGAuc	UCCAGGGCGGAGACUUCAC	-2.4	-2.2	-1.9	1	-1	6	-44.4	42.5	-1	2	II	63.2	37.1	41.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2686	M60857	Reynolds	Cyclophilin B	829639	CUUCAGUUUGAAGUUCUCAuc	UGAGAACUUCAAACUGAAG	-2.1	-2.1	-2.1	1	-1	1	-33.2	69.5	1.6	4	II	36.8	51.1	41.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2687	M60857	Reynolds	Cyclophilin B	829639	GCCACAAAAUUAUCCACUGuu	CAGUGGAUAAUUUUGUGGC	-2.1	-3.4	2	0	0	3	-34.4	58.6	0	2	II	42.1	42.8	39.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2688	M60857	Reynolds	Cyclophilin B	829639	GCUAAGGCCACAAAAUUAUcc	AUAAUUUUGUGGCCUUAGC	-1.1	-3.4	1.2	1	-2	4	-33.3	56.5	-6.2	6	III	36.8	38.6	32.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2689	M60857	Reynolds	Cyclophilin B	829639	CUUUCUCUCCUGUAGCUAAgg	UUAGCUACAGGAGAGAAAG	-0.9	-2.1	1.6	0	-3	2	-36.1	72.7	-5.8	4	II	42.1	39.5	30.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2690	M60857	Reynolds	Cyclophilin B	829639	CCUGUGCCAUCUCCCCUGGug	CCAGGGGAGAUGGCACAGG	-3.3	-3.3	-1	-1	0	4	-46	43.3	1.1	1	II	68.4	39	21.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2691	M60857	Reynolds	Cyclophilin B	829639	CCGUAGUGCUUCAGUUUGAag	UCAAACUGAAGCACUACGG	-2.4	-3.3	0.5	0	-4	3	-37.1	47.5	-8	4	III	47.4	32.9	18.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2692	M60857	Reynolds	Cyclophilin B	829639	UGCCAUCUCCCCUGGUGAAgu	UUCACCAGGGGAGAUGGCA	-0.9	-2.1	-1.5	1	-1	4	-43.9	48.8	-5.8	2	II	57.9	35.9	17.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2693	M60857	Reynolds	Cyclophilin B	829639	CACAAAAUUAUCCACUGUUuu	AACAGUGGAUAAUUUUGUG	-0.9	-2.1	-0.1	-1	-2	2	-30.8	70.1	-3.2	5	II	31.6	50.7	16.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2694	M60857	Reynolds	Cyclophilin B	829639	CCUGUAGCUAAGGCCACAAaa	UUGUGGCCUUAGCUACAGG	-0.9	-3.3	-7.1	-1	-3	4	-40.2	44	-3.5	3	III	52.6	32.5	16	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2695	M60857	Reynolds	Cyclophilin B	829639	GCUGUUUUUGUAGCCAAAUcc	AUUUGGCUACAAAAACAGC	-1.1	-3.4	-1.3	1	-2	3	-32.8	63.4	-4	2	II	36.8	41.5	13.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2696	M60857	Reynolds	Cyclophilin B	829639	GAUGGAAUUUGCUGUUUUUgu	AAAAACAGCAAAUUCCAUC	-0.9	-2.4	4.5	-1	-1	2	-30.6	70.2	-6.3	4	II	31.6	38.5	7.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2697	M60857	Reynolds	Cyclophilin B	829639	ACGAUGGAAUUUGCUGUUUuu	AAACAGCAAAUUCCAUCGU	-0.9	-2.2	2.7	1	-1	2	-33.4	63.7	-1	6	II	36.8	47.3	7.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2698	M60857	Reynolds	Cyclophilin B	829639	GGGAAGCGCUCACCGUAGAug	UCUACGGUGAGCGCUUCCC	-2.4	-3.3	-0.1	1	-3	4	-43.7	31.4	-1	1	III	63.2	23.1	2.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2699	M60857	Reynolds	Cyclophilin B	829639	CGGGGAAGCGCUCACCGUAga	UACGGUGAGCGCUUCCCCG	-1.3	-2.4	-0.6	0	-4	5	-44.9	24.4	-2.7	0	III	68.4	28.7	-9.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2700	-	Reynolds	RenillaLuc	/	CACAACAUGUCGCCAUAAAua	UUUAUGGCGACAUGUUGUG	-0.9	-2.1	-2.8	0	-4	4	-34.8	58.4	-3	3	II	42.1	32.3	29.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2701	-	Reynolds	RenillaLuc	/	UUCAAACCAUGCAGUAAGAua	UCUUACUGCAUGGUUUGAA	-2.4	-0.9	0.8	-1	0	2	-34.5	64	-6	6	II	36.8	50.5	52.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2702	-	Reynolds	RenillaLuc	/	AAUCACAUCUACUACACUUuc	AAGUGUAGUAGAUGUGAUU	-0.9	-0.9	3.3	3	2	1	-32.8	80.1	-1.3	6	II	31.6	61.2	87.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2703	-	Reynolds	RenillaLuc	/	GUUCUAACUUUCUCAUGAUuu	AUCAUGAGAAAGUUAGAAC	-1.1	-2.2	1.1	1	0	1	-31.6	70	-6.3	4	II	31.6	40.4	47.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2704	-	Reynolds	RenillaLuc	/	CAAGAUAUGCUGCAAAUUCuu	GAAUUUGCAGCAUAUCUUG	-2.4	-2.1	2.2	0	0	2	-32.4	66.3	10.5	5	II	36.8	60.8	84.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2705	-	Reynolds	RenillaLuc	/	UAAUGUUGGACGACGAACUuc	AGUUCGUCGUCCAACAUUA	-2.1	-1.3	0.5	2	2	2	-35.5	75.5	-8.9	8	II	42.1	74	66.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2706	-	Reynolds	RenillaLuc	/	AUCAUCACUUGCACGUAGAua	UCUACGUGCAAGUGAUGAU	-2.4	-1.1	-0.4	1	0	2	-36.6	70.2	-8.7	6	II	42.1	52.7	89.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2707	-	Reynolds	RenillaLuc	/	CAUUUCAUCAGGUGCAUCUuc	AGAUGCACCUGAUGAAAUG	-2.1	-2.1	1.6	-1	-1	2	-35.9	81	-6	7	II	42.1	54.3	63.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2708	-	Reynolds	RenillaLuc	/	UCAUCCGUUUCCUUUGUUCug	GAACAAAGGAAACGGAUGA	-2.4	-2.4	2.6	0	3	3	-34.9	84.9	5	7	II	42.1	65.2	84.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2709	-	Reynolds	RenillaLuc	/	UCCUUCUUCAGAUUUGAUCaa	GAUCAAAUCUGAAGAAGGA	-2.4	-2.4	0	2	2	2	-34	96.7	9.7	7	Ib	36.8	70.1	90.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2710	-	Reynolds	RenillaLuc	/	UAUUAGGAAACUUCUUGGCac	GCCAAGAAGUUUCCUAAUA	-3.4	-1.3	1.8	0	5	3	-33.6	88.5	10.8	9	Ia	36.8	74.5	72.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2711	-	Reynolds	RenillaLuc	/	AAUGGUUCAAGAUAUGCUGca	CAGCAUAUCUUGAACCAUU	-2.1	-0.9	1.1	2	4	2	-33.6	79.8	0	6	Ia	36.8	69	83.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2712	NM_031313	Reynolds	SEAP	7124	AGAUGAUGAGGUUCUUGGCgg	GCCAAGAACCUCAUCAUCU	-3.4	-2.1	0.8	2	4	3	-38.4	70.8	10.4	5	Ia	47.4	56.9	90.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2713	NM_031313	Reynolds	SEAP	7124	CUGUAGUCAUCUGGGUACUca	AGUACCCAGAUGACUACAG	-2.1	-2.1	0.9	-1	-2	3	-38.6	57	-2.6	4	II	47.4	42.7	1.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2714	NM_031313	Reynolds	SEAP	7124	GUAUAGGAGGACCGUGUAGgc	CUACACGGUCCUCCUAUAC	-2.1	-2.2	-0.7	0	1	3	-39.3	48.9	5.1	6	II	52.6	41.8	21.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2715	NM_031313	Reynolds	SEAP	7124	AUAGCCUGGACCGUUUCCGua	CGGAAACGGUCCAGGCUAU	-2.4	-1.1	-3.5	3	6	3	-41.1	61.7	-2.4	4	Ib	57.9	67.7	80.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2716	NM_031313	Reynolds	SEAP	7124	ACAUAGCCUGGACCGUUUCcg	GAAACGGUCCAGGCUAUGU	-2.4	-2.2	-1.2	1	2	3	-39.7	56.7	12.5	5	Ib	52.6	52.9	6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2717	NM_031313	Reynolds	SEAP	7124	CAUGACGUGCGCUAUGAAGgu	CUUCAUAGCGCACGUCAUG	-2.1	-2.1	1.2	0	0	4	-37.8	56.9	-1.7	4	II	52.6	45.1	16.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2718	NM_031313	Reynolds	SEAP	7124	UAUAGGAGGACCGUGUAGGcc	CCUACACGGUCCUCCUAUA	-3.3	-1.3	-0.7	1	5	3	-40.4	64.7	-2.4	8	Ia	52.6	72	62.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2719	U47296	Reynolds	FireflyLuc	116160065	UAUCAUGUCUGCUCGAAGCgg	GCUUCGAGCAGACAUGAUA	-3.4	-1.3	-1.2	2	4	2	-38	79.1	7.8	6	Ia	47.4	73.2	97.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2720	U47296	Reynolds	FireflyLuc	116160065	UUUUAGAAUCCAUGAUAAUaa	AUUAUCAUGGAUUCUAAAA	-1.1	-0.9	2	-1	1	2	-28.3	81.8	-3.6	8	II	21.1	60.4	55.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2721	U47296	Reynolds	FireflyLuc	116160065	CAAAUAUCCGAGUGUAGUAaa	UACUACACUCGGAUAUUUG	-1.3	-2.1	1.4	0	-2	3	-33.4	68.6	-3.5	5	II	36.8	43.9	38.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2722	U47296	Reynolds	FireflyLuc	116160065	GGAUUGUUUACAUAACCGGac	CCGGUUAUGUAAACAAUCC	-3.3	-3.3	0.3	1	2	4	-33.9	67.5	-2.4	5	II	42.1	55.1	41.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2723	U47296	Reynolds	FireflyLuc	116160065	CGCUUCCGGAUUGUUUACAua	UGUAAACAAUCCGGAAGCG	-2.1	-2.4	0	0	-3	4	-36.3	61.8	-0.4	3	III	47.4	34.4	9.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2724	U47296	Reynolds	FireflyLuc	116160065	GGCGUUGGUCGCUUCCGGAuu	UCCGGAAGCGACCAACGCC	-2.4	-3.3	-2.3	0	-1	4	-44.8	39.5	-3.8	0	III	68.4	28.8	89.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2725	U47296	Reynolds	FireflyLuc	116160065	GUACUUAAUCAGAGACUUCag	GAAGUCUCUGAUUAAGUAC	-2.4	-2.2	0.8	0	1	1	-33.1	64.9	7.1	5	II	36.8	54.6	34.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2726	U47296	Reynolds	FireflyLuc	116160065	UUUGUAUUCAGCCCAUAUCgu	GAUAUGGGCUGAAUACAAA	-2.4	-0.9	1.3	2	3	4	-34	89.8	7.5	8	Ia	36.8	76.9	91.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2727	U47296	Reynolds	FireflyLuc	116160065	UACUUAAUCAGAGACUUCAgg	UGAAGUCUCUGAUUAAGUA	-2.1	-1.3	0.8	1	0	1	-33	84.3	-3.8	9	II	31.6	67.5	91.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2728	NM_020548	Reynolds	DBI	1622	GCUCAUUCCAGGCAUCCCAcu	UGGGAUGCCUGGAAUGAGC	-2.1	-3.4	-0.3	3	-1	3	-42.9	50.6	1.4	3	II	57.9	39.4	4.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2729	NM_020548	Reynolds	DBI	1622	GUAAGCUUUCAUGGCAUCUuc	AGAUGCCAUGAAAGCUUAC	-2.1	-2.2	0.3	0	0	3	-35.9	67.9	1.4	5	II	42.1	49.9	-9.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2730	NM_020548	Reynolds	DBI	1622	AACAAGGCAUUGUCUCAGUuu	ACUGAGACAAUGCCUUGUU	-2.2	-0.9	-1	1	1	3	-36.6	64	-1.3	5	II	42.1	53.3	6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2731	NM_020548	Reynolds	DBI	1622	GGAAAGUCAGUAAUCGUUUua	AAACGAUUACUGACUUUCC	-0.9	-3.3	1.9	1	-1	2	-32.7	61.9	-1.5	5	II	36.8	46.5	-18.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2732	NM_020548	Reynolds	DBI	1622	ACUCAAGGAAAGUCAGUAAuc	UUACUGACUUUCCUUGAGU	-0.9	-2.2	-0.7	2	-1	2	-34.6	61.7	1.3	7	II	36.8	39.7	-8.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2733	NM_020548	Reynolds	DBI	1622	UAAGCUUUCAUGGCAUCUUcc	AAGAUGCCAUGAAAGCUUA	-0.9	-1.3	0.3	2	2	3	-34.6	78.8	-4	7	II	36.8	70.8	50.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2734	NM_020548	Reynolds	DBI	1622	UUGUUGAUGUAAGCUUUCAug	UGAAAGCUUACAUCAACAA	-2.1	-0.9	1.3	1	0	2	-32.1	82.1	-1.5	8	II	31.6	64.7	33.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2735	NM_020548	Reynolds	DBI	1622	UCUUUAGCUCUUCUACUUUgu	AAAGUAGAAGAGCUAAAGA	-0.9	-2.4	-0.2	0	1	2	-32.5	83.9	3.8	8	II	31.6	58.9	71.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2736	NM_020548	Reynolds	DBI	1622	UAUUUUUUCUUUAGCUCUUcu	AAGAGCUAAAGAAAAAAUA	-0.9	-1.3	5.2	1	2	2	-27.5	100.6	-3.5	8	II	21.1	70.9	68	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2737	X75932	Reynolds	PLK	5347	GCACUUGGCAAAGCCGCCCuu	GGGCGGCUUUGCCAAGUGC	-3.3	-3.4	-4.2	2	2	7	-44.6	38.9	9.7	2	II	68.4	50.3	30.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2738	X75932	Reynolds	PLK	5347	GGAUAUUUCCAUGGACAUCuu	GAUGUCCAUGGAAAUAUCC	-2.4	-3.3	-1.3	1	0	2	-35.5	65.8	15.4	3	II	42.1	46.1	43.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2739	X75932	Reynolds	PLK	5347	CACAUCCACCUCGAAACUGug	CAGUUUCGAGGUGGAUGUG	-2.1	-2.1	1.1	0	0	2	-38.3	50	-4.4	3	II	52.6	47.6	52.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2740	X75932	Reynolds	PLK	5347	GGAAUACUGUAUUCAUUCUuc	AGAAUGAAUACAGUAUUCC	-2.1	-3.3	-0.8	1	-1	2	-31.9	77.9	1.4	4	II	31.6	40	58	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2741	X75932	Reynolds	PLK	5347	AUGAGGCGUGUUGAGUCAUug	AUGACUCAACACGCCUCAU	-1.1	-1.1	1.3	1	1	4	-38.6	50.4	-1	4	II	47.4	48	58.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2742	X75932	Reynolds	PLK	5347	UCAUGUAAUUGCGGAAAUAuu	UAUUUCCGCAAUUACAUGA	-1.3	-2.4	1.7	-1	-1	4	-31.9	63	-6.1	6	II	31.6	51.2	57.1	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2743	X75932	Reynolds	PLK	5347	AGAGACUUAGGCACAAUCUug	AGAUUGUGCCUAAGUCUCU	-2.1	-2.1	-1.6	2	1	3	-37.2	65.5	-6.2	5	II	42.1	58.2	68.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2744	X75932	Reynolds	PLK	5347	UUCACCUCCAGAUCUUCAUuc	AUGAAGAUCUGGAGGUGAA	-1.1	-0.9	0.4	2	1	2	-37.4	86.7	-4	7	II	42.1	62.8	55	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2745	X75932	Reynolds	PLK	5347	UUAGGCAAGAAGUCUCAAAag	UUUGAGACUUCUUGCCUAA	-0.9	-0.9	-0.3	1	0	3	-34.5	72.4	-1.5	7	II	36.8	60.5	82.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2746	X75932	Reynolds	PLK	5347	UAUUUAAGGAGGGUGAUCUuc	AGAUCACCCUCCUUAAAUA	-2.1	-1.3	1.7	1	3	3	-35.2	85.2	-1.6	10	II	36.8	70.6	72.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2747	NM_002046	Reynolds	GAPDH	2597	CCCUUUUGGCUCCCCCCUGca	CAGGGGGGAGCCAAAAGGG	-2.1	-3.3	-2.3	1	-1	6	-45.4	45.5	-2	1	II	68.4	43.8	-24.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2748	NM_002046	Reynolds	GAPDH	2597	AGGGGGCAGAGAUGAUGACcc	GUCAUCAUCUCUGCCCCCU	-2.2	-2.1	2.7	1	2	6	-43.2	37.5	7.4	2	II	57.9	41.5	69.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2749	NM_002046	Reynolds	GAPDH	2597	GAACAUGGGGGCAUCAGCAga	UGCUGAUGCCCCCAUGUUC	-2.1	-2.4	-0.6	2	1	6	-42.7	42.1	-3.7	3	II	57.9	43.2	12.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2750	NM_002046	Reynolds	GAPDH	2597	UCAUGGUUCACACCCAUGAcg	UCAUGGGUGUGAACCAUGA	-2.4	-2.4	-7	2	0	3	-39.4	67	-3.7	7	II	47.4	61.6	60.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2751	NM_002046	Reynolds	GAPDH	2597	UGAGGCUGUUGUCAUACUUcu	AAGUAUGACAACAGCCUCA	-0.9	-2.1	0.8	1	1	3	-37	67.8	1.5	5	II	42.1	57.6	94.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2752	NM_002046	Reynolds	GAPDH	2597	GGAGGCAUUGCUGAUGAUCuu	GAUCAUCAGCAAUGCCUCC	-2.4	-3.3	0.4	0	0	3	-39.8	48.2	8.1	2	II	52.6	44.6	70.8	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2753	NM_002046	Reynolds	GAPDH	2597	GCUAAGCAGUUGGUGGUGCag	GCACCACCAACUGCUUAGC	-3.4	-3.4	-1.3	1	2	2	-41.2	43.7	12.4	2	II	57.9	38.9	38.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2754	NM_002046	Reynolds	GAPDH	2597	GGAUGACCUUGGCCAGGGGug	CCCCUGGCCAAGGUCAUCC	-3.3	-3.3	-3.6	0	2	4	-46	33.4	2.4	0	II	68.4	30.5	-2.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2755	NM_002046	Reynolds	GAPDH	2597	CACGAUACCAAAGUUGUCAug	UGACAACUUUGGUAUCGUG	-2.1	-2.1	1.1	0	-2	2	-34.9	66.2	1.6	5	II	42.1	49.9	43	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2756	NM_002046	Reynolds	GAPDH	2597	ACUGUGGUCAUGAGUCCUUcc	AAGGACUCAUGACCACAGU	-0.9	-2.2	-0.7	1	1	2	-39.3	51.3	-6.6	6	II	47.4	50.3	-17.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2757	NM_002046	Reynolds	GAPDH	2597	UGAUGACCCUUUUGGCUCCcc	GGAGCCAAAAGGGUCAUCA	-3.3	-2.1	-1.5	0	3	3	-40.3	64.3	10.4	5	Ib	52.6	61.6	88.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2758	NM_002046	Reynolds	GAPDH	2597	GAGAUGAUGACCCUUUUGGcu	CCAAAAGGGUCAUCAUCUC	-3.3	-2.4	-0.6	1	2	3	-37.1	72.4	0	6	II	47.4	55.9	93.9	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2759	NM_002046	Reynolds	GAPDH	2597	UUCACACCCAUGACGAACAug	UGUUCGUCAUGGGUGUGAA	-2.1	-0.9	-1.5	1	0	3	-38.4	66.6	-6.3	5	II	47.4	62	80.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2760	NM_002046	Reynolds	GAPDH	2597	UACUUCUCAUGGUUCACACcc	GUGUGAACCAUGAGAAGUA	-2.2	-1.3	0.9	1	4	2	-36.1	92	9.7	7	Ia	42.1	68.6	92.7	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2761	NM_002046	Reynolds	GAPDH	2597	UUGUCAUACUUCUCAUGGUuc	ACCAUGAGAAGUAUGACAA	-2.2	-0.9	-0.3	2	2	2	-35	80.6	-6.3	6	II	36.8	63.6	92.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2762	NM_002046	Reynolds	GAPDH	2597	UGUUGUCAUACUUCUCAUGgu	CAUGAGAAGUAUGACAACA	-2.1	-2.1	0	0	2	1	-33.8	86.5	0.7	6	Ib	36.8	65.5	92.4	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2763	NM_002046	Reynolds	GAPDH	2597	CUUGAGGCUGUUGUCAUACuu	GUAUGACAACAGCCUCAAG	-2.2	-2.1	0.3	-1	1	3	-37	60.9	13.5	5	II	47.4	55.1	77.6	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2764	NM_002046	Reynolds	GAPDH	2597	AUCUUGAGGCUGUUGUCAUac	AUGACAACAGCCUCAAGAU	-1.1	-1.1	0.3	2	2	3	-37	71.3	-1.6	6	II	42.1	52.3	58.3	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2765	NM_002046	Reynolds	GAPDH	2597	UGAUCUUGAGGCUGUUGUCau	GACAACAGCCUCAAGAUCA	-2.4	-2.1	1.3	0	3	3	-38.3	76.1	5.1	6	Ia	47.4	62.6	94.2	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2766	NM_002046	Reynolds	GAPDH	2597	UCAUGGAUGACCUUGGCCAgg	UGGCCAAGGUCAUCCAUGA	-2.1	-2.4	-3.5	1	2	4	-41.7	63	-6.1	6	II	52.6	61.3	91.5	branched DNA (bDNA) technology	HEK293	100nM	24h	Rational siRNA design for RNA interference.	14758366	2004	Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.	Nature Biotechnology	2004/2/1
Si2767	U92436	Vickers	MMAC1	5728	GCGAGAGGCGGACGGGACCgc	GGUCCCGUCCGCCUCUCGC	-3.3	-3.4	-1.2	0	0	5	-48.7	19.1	9.8	0	II	78.9	34.4	0	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2768	U92436	Vickers	MMAC1	5728	CCGGGCGCCUCGGAAGACCga	GGUCUUCCGAGGCGCCCGG	-3.3	-3.3	-2.7	0	-1	9	-48.3	18.1	3.1	-1	II	78.9	38.5	0	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2769	U92436	Vickers	MMAC1	5728	AUGGCUGCAGCUUCCGAGAgg	UCUCGGAAGCUGCAGCCAU	-2.4	-1.1	-0.5	1	0	3	-43.1	56.6	-3.1	3	II	57.9	50.7	40	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2770	U92436	Vickers	MMAC1	5728	CCCCGCGGCUGCUCACAGGcg	CCUGUGAGCAGCCGCGGGG	-3.3	-3.3	-8.1	1	-1	9	-48.7	26.2	-4.4	-1	II	78.9	32.8	25	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2771	U92436	Vickers	MMAC1	5728	UCAGGAGAAGCCGAGGAAGag	CUUCCUCGGCUUCUCCUGA	-2.1	-2.4	-0.8	0	2	4	-42.1	36.4	-4.7	4	II	57.9	53.5	62	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2772	U92436	Vickers	MMAC1	5728	CGGGAGGUGCCGCCGCCGCcg	GCGGCGGCGGCACCUCCCG	-3.4	-2.4	-1.2	0	0	11	-51.8	14.6	8.2	0	II	89.5	37.5	20	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2773	U92436	Vickers	MMAC1	5728	CCCGGGUCCCUGGAUGUGCcg	GCACAUCCAGGGACCCGGG	-3.4	-3.3	-0.6	1	-1	6	-47.6	29.1	5.4	1	II	73.7	41.8	32	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2774	U92436	Vickers	MMAC1	5728	UCCUCCGAACGGCUGCCUCcg	GAGGCAGCCGUUCGGAGGA	-2.4	-2.4	-1	0	3	4	-45.9	55.2	12.1	3	II	68.4	53.9	58	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2775	U92436	Vickers	MMAC1	5728	UUCUCCUCAGCAGCCAGAGgc	CUCUGGCUGCUGAGGAGAA	-2.1	-0.9	-2.2	1	3	3	-42.8	63.1	-2.4	4	Ib	57.9	61.9	52	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2776	U92436	Vickers	MMAC1	5728	CCGCUUGGCUCUGGACCGCag	GCGGUCCAGAGCCAAGCGG	-3.4	-3.3	-2.4	0	0	4	-46.6	26.1	6.1	0	II	73.7	36.5	31	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2777	U92436	Vickers	MMAC1	5728	UUCUUCUGCAGGAUGGAAAug	UUUCCAUCCUGCAGAAGAA	-0.9	-0.9	0.7	2	0	2	-36.9	82	-4	8	II	42.1	58.8	27	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2778	U92436	Vickers	MMAC1	5728	UGGAUAAAUAUAGGUCAAGuc	CUUGACCUAUAUUUAUCCA	-2.1	-2.1	3	-1	1	2	-31.9	67.6	0	7	Ia	31.6	62.5	48	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2779	U92436	Vickers	MMAC1	5728	AUCAAUAUUGUUCCUGUAUac	AUACAGGAACAAUAUUGAU	-1.1	-1.1	0.4	1	0	2	-30.6	88.8	4.2	7	II	26.3	59.2	49	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2780	U92436	Vickers	MMAC1	5728	AUUAAAUUUGGCGGUGUCAua	UGACACCGCCAAAUUUAAU	-2.1	-1.1	2	1	1	5	-33.5	70.1	-3.4	8	II	36.8	54.9	65	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2781	U92436	Vickers	MMAC1	5728	UCAAGAUCUUCACAAAAGGgu	CCUUUUGUGAAGAUCUUGA	-3.3	-2.4	0.5	1	3	2	-33.3	66.5	0.1	6	Ia	36.8	67.5	73	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2782	U92436	Vickers	MMAC1	5728	CAUUACACCAGUUCGUCCCuu	GGGACGAACUGGUGUAAUG	-3.3	-2.1	0.7	0	2	3	-38.5	78.6	11.1	6	II	52.6	61.8	75	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2783	U92436	Vickers	MMAC1	5728	UUGUCUCUGGUCCUUACUUcc	AAGUAAGGACCAGAGACAA	-0.9	-0.9	1.8	1	2	2	-37	79.6	-1.6	5	II	42.1	63.4	75	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2784	U92436	Vickers	MMAC1	5728	CACAUAGCGCCUCUGACUGgg	CAGUCAGAGGCGCUAUGUG	-2.1	-2.1	-0.5	0	1	5	-40.9	57.2	3.4	3	II	57.9	47.6	73	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2785	U92436	Vickers	MMAC1	5728	GGAAUAUAUCUUCACCUUUag	AAAGGUGAAGAUAUAUUCC	-0.9	-3.3	0.2	1	-2	2	-31.8	71.5	1.5	5	II	31.6	45.6	9	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2786	U92436	Vickers	MMAC1	5728	UGGAAGAACUCUACUUUGAua	UCAAAGUAGAGUUCUUCCA	-2.4	-2.1	1.2	1	-1	2	-34.8	71	-3	7	II	36.8	56.2	30	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2787	U92436	Vickers	MMAC1	5728	AUGAAGAAUGUAUUUACCCaa	GGGUAAAUACAUUCUUCAU	-3.3	-1.1	-0.1	1	5	3	-31.6	80	12.4	6	Ia	31.6	78.2	70	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2788	U92436	Vickers	MMAC1	5728	CGGUUGGCUUUGUCUUUAUuu	AUAAAGACAAAGCCAACCG	-1.1	-2.4		-1	-4	3	-34.5	64.5	-6	4	III	42.1	34.8	0	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2789	U92436	Vickers	MMAC1	5728	CUGCUAGCCUCUGGAUUUGac	CAAAUCCAGAGGCUAGCAG	-2.1	-2.1	1.9	0	-1	3	-39.2	48	-1.3	4	II	52.6	41.4	44	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2790	U92436	Vickers	MMAC1	5728	CUCUGGAUCAGAGUCAGUGgu	CACUGACUCUGAUCCAGAG	-2.1	-2.1	-3.5	0	1	2	-39.7	59.1	0.3	3	II	52.6	54.2	15	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2791	U92436	Vickers	MMAC1	5728	UUAUUUUCAUGGUGUUUUAuc	UAAAACACCAUGAAAAUAA	-1.3	-0.9	1.1	0	0	2	-27.5	99.9	-3.8	9	II	21.1	62.7	58	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2792	U92436	Vickers	MMAC1	5728	UUGUUCCUAUAACUGGUAAuc	UUACCAGUUAUAGGAACAA	-0.9	-0.9	-0.5	0	0	2	-32.6	87.9	3.3	7	II	31.6	52.9	40	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2793	U92436	Vickers	MMAC1	5728	AGUGUCAAAACCCUGUGGAug	UCCACAGGGUUUUGACACU	-2.4	-2.1	0.2	1	2	3	-38.8	64.2	-3.7	6	II	47.4	48.3	45	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2794	U92436	Vickers	MMAC1	5728	AACUGGAAUAAAACGGGAAag	UUCCCGUUUUAUUCCAGUU	-0.9	-0.9	3.8	1	0	4	-33.5	63	-4	6	II	36.8	49.4	38	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2795	U92436	Vickers	MMAC1	5728	CACUUCAGUUGGUGACAGAac	UCUGUCACCAACUGAAGUG	-2.4	-2.1	0.2	-2	-3	2	-37.9	56.6	-5.8	4	II	47.4	33.9	25	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2796	U92436	Vickers	MMAC1	5728	GUAGCAAAACCUUUCGGAAac	UUCCGAAAGGUUUUGCUAC	-0.9	-2.2	-3.6	0	0	3	-34.6	61.4	-0.8	2	II	42.1	35.9	32	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2797	U92436	Vickers	MMAC1	5728	AAAUUAUUUCCUUUCUGAGca	CUCAGAAAGGAAAUAAUUU	-2.1	-0.9	1.2	2	5	2	-28.7	98.8	3	7	Ia	26.3	66	30	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2798	U92436	Vickers	MMAC1	5728	GUAAAUAGCUGGAGAUGGUau	ACCAUCUCCAGCUAUUUAC	-2.2	-2.2	4.3	0	1	2	-36.3	54.6	-3.9	5	II	42.1	42.9	6	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2799	U92436	Vickers	MMAC1	5728	CCAGAUUAAUAACUGUAGCau	GCUACAGUUAUUAAUCUGG	-3.4	-3.3	-1.9	0	1	2	-32.8	70.6	15.1	2	II	36.8	53.3	37	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2800	U92436	Vickers	MMAC1	5728	ACCCCAAUACAGAUUCACUuc	AGUGAAUCUGUAUUGGGGU	-2.1	-2.2	0.3	2	1	4	-37.1	64.9	-1.9	3	II	42.1	49.2	39	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2801	U92436	Vickers	MMAC1	5728	CAUUGUUGCUGUGUUUCUUac	AAGAAACACAGCAACAAUG	-0.9	-2.1	2.2	1	0	2	-32.7	80.4	-0.6	4	II	36.8	46.5	30	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2802	U92436	Vickers	MMAC1	5728	AUGUUUCAAGCCCAUUCUUug	AAGAAUGGGCUUGAAACAU	-0.9	-1.1	0.9	1	1	4	-34.2	73.1	-6.3	5	II	36.8	54.2	40	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2803	J03132	Vickers	ICAM-1	3383	GAGGAGCUCAGCGUCGACUgg	AGUCGACGCUGAGCUCCUC	-2.1	-2.4	0.7	1	0	3	-43.9	43.2	-1.3	2	III	63.2	42	0	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2804	J03132	Vickers	ICAM-1	3383	GCUGAGGUUGCAACUCUGAgu	UCAGAGUUGCAACCUCAGC	-2.4	-3.4	-1.5	0	-2	2	-40.3	39.4	-6	3	III	52.6	33.4	1	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2805	J03132	Vickers	ICAM-1	3383	CAGGCAGGAGCAACUCCUUuu	AAGGAGUUGCUCCUGCCUG	-0.9	-2.1	-4.7	-1	-2	3	-42.3	35.6	-10.9	0	III	57.9	38	-15	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2806	J03132	Vickers	ICAM-1	3383	UGAAUAGCACAUUGGUUGGcu	CCAACCAAUGUGCUAUUCA	-3.3	-2.1	2.1	-1	3	2	-35.6	77.3	5.3	8	Ia	42.1	63.4	7	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2807	J03132	Vickers	ICAM-1	3383	CCCACUGGCUGCCAAGAGGgg	CCUCUUGGCAGCCAGUGGG	-3.3	-3.3	-0.5	0	-1	3	-45.6	40.1	-2	1	II	68.4	38.7	18	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2808	J03132	Vickers	ICAM-1	3383	CUCUCCUCACCAGCACCGUgg	ACGGUGCUGGUGAGGAGAG	-2.2	-2.1	1.3	0	-1	3	-44.2	45.8	-8.3	2	III	63.2	39.3	-20	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2809	J03132	Vickers	ICAM-1	3383	AAGGUCUGGAGCUGGUAGGgg	CCUACCAGCUCCAGACCUU	-3.3	-0.9	2.6	2	3	2	-42.7	60.2	-2.4	5	Ib	57.9	59.2	45	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2810	J03132	Vickers	ICAM-1	3383	CGUGUCCACCUCUAGGACCcg	GGUCCUAGAGGUGGACACG	-3.3	-2.4	-5.7	0	0	2	-43.4	49.1	3.1	0	II	63.2	47.8	10	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2811	J03132	Vickers	ICAM-1	3383	CAGUGCCAGGUGGACCUGGgc	CCAGGUCCACCUGGCACUG	-3.3	-2.1	-7.5	0	0	3	-45.7	34	-2	1	II	68.4	44.4	21	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2812	J03132	Vickers	ICAM-1	3383	CAGUAUUACUGCACACGUCag	GACGUGUGCAGUAAUACUG	-2.4	-2.1	-2.1	1	0	2	-36.4	63.7	3.2	3	II	47.4	53.8	30	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2813	J03132	Vickers	ICAM-1	3383	CUCUGGCUUCGUCAGAAUCac	GAUUCUGACGAAGCCAGAG	-2.4	-2.1	-2.7	-1	-1	3	-38.8	60.7	10.8	1	II	52.6	47.2	-10	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2814	J03132	Vickers	ICAM-1	3383	GUGGCCUUCAGCAGGAGCUgg	AGCUCCUGCUGAAGGCCAC	-2.1	-2.2	-2.1	1	0	4	-44.8	49.1	-8.9	1	III	63.2	44.6	10	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2815	J03132	Vickers	ICAM-1	3383	AUACAGGACACGAAGCUCCcg	GGAGCUUCGUGUCCUGUAU	-3.3	-1.1	-1.5	3	4	2	-40	54.4	5.2	7	Ib	52.6	72.5	58	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2816	J03132	Vickers	ICAM-1	3383	AUCCUUUAGACACUUGAGCuc	GCUCAAGUGUCUAAAGGAU	-3.4	-1.1	0.3	4	4	2	-36	78.5	9.8	6	Ia	42.1	69	40	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2817	J03132	Vickers	ICAM-1	3383	CUCCUGGCCCGACAGAGGUag	ACCUCUGUCGGGCCAGGAG	-2.2	-2.1	-1.7	1	-1	6	-46.5	35.7	-3.6	1	III	68.4	39.7	5	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2818	J03132	Vickers	ICAM-1	3383	CUACCACAGUGAUGAUGACaa	GUCAUCAUCACUGUGGUAG	-2.2	-2.1	-1.3	0	1	2	-37.5	48.3	5.4	1	II	47.4	36.4	35	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2819	J03132	Vickers	ICAM-1	3383	UGUGUGUUCGGUUUCAUGGgg	CCAUGAAACCGAACACACA	-3.3	-2.1	2.1	2	4	3	-36.9	87.5	5.4	7	Ib	47.4	73.5	59	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2820	J03132	Vickers	ICAM-1	3383	GAGGCGUGGCUUGUGUGUUcg	AACACACAAGCCACGCCUC	-0.9	-2.4	1.3	2	0	4	-41.4	46.7	1.4	1	III	57.9	36.9	1	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2821	J03132	Vickers	ICAM-1	3383	CUGUCCCGGGAUAGGUUCAgg	UGAACCUAUCCCGGGACAG	-2.1	-2.1	-1.7	-1	-3	6	-42.2	50.9	-3	1	III	57.9	34.5	20	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2822	J03132	Vickers	ICAM-1	3383	GAGGAAGAGGCCCUGUCCCgg	GGGACAGGGCCUCUUCCUC	-3.3	-2.4	-3.6	1	3	5	-46.4	37	10.1	1	II	68.4	47.3	52	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2823	J03132	Vickers	ICAM-1	3383	CCACUCUGUUCAGUGUGGCac	GCCACACUGAACAGAGUGG	-3.4	-3.3	-3.5	1	2	3	-41.3	48.3	5.4	1	II	57.9	45.1	21	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2824	J03132	Vickers	ICAM-1	3383	CUGACUGAGGACAAUGCCCug	GGGCAUUGUCCUCAGUCAG	-3.3	-2.1	-4.1	0	1	4	-41.6	53.9	2.8	1	II	57.9	59.3	62	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2825	J03132	Vickers	ICAM-1	3383	AGGUGUGCAGGUACCAUGGcc	CCAUGGUACCUGCACACCU	-3.3	-2.1	-1.4	1	2	2	-42.5	59.1	0.4	3	II	57.9	57.3	75	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2826	J03132	Vickers	ICAM-1	3383	CUCUCAUCAGGCUAGACUUua	AAGUCUAGCCUGAUGAGAG	-0.9	-2.1	-0.9	0	-2	3	-38.6	66.7	-8.3	3	II	47.4	44.7	50	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2827	J03132	Vickers	ICAM-1	3383	CAGUUGUAUGUCCUCAUGGug	CCAUGAGGACAUACAACUG	-3.3	-2.1	-0.3	0	1	2	-37.1	71	0.4	5	II	47.4	60.7	58	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2828	J03132	Vickers	ICAM-1	3383	GGCCUCAGCAUACCCAAUAgg	UAUUGGGUAUGCUGAGGCC	-1.3	-3.3	-0.8	2	-4	4	-40.8	47.3	-1.4	3	III	52.6	34	43	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2829	J03132	Vickers	ICAM-1	3383	UGCUACACAUGUCUAUGGAgg	UCCAUAGACAUGUGUAGCA	-2.4	-2.1	-0.9	0	1	2	-37.6	74.3	6.7	6	II	42.1	48	63	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2830	J03132	Vickers	ICAM-1	3383	CCCAAGCUGGCAUCCGUCAgg	UGACGGAUGCCAGCUUGGG	-2.1	-3.3	-1	0	-4	3	-44	38	-5.4	1	III	63.2	31.9	29	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2831	J03132	Vickers	ICAM-1	3383	GUGCCCAAGCUGGCAUCCGuc	CGGAUGCCAGCUUGGGCAC	-2.4	-2.2	-5.1	2	2	4	-45	35.2	0	0	II	68.4	45.4	24	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2832	J03132	Vickers	ICAM-1	3383	AGUGCCCAAGCUGGCAUCCgu	GGAUGCCAGCUUGGGCACU	-3.3	-2.1	-5.1	0	3	4	-44.7	47.7	8.1	2	II	63.2	50.5	40	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2833	J03132	Vickers	ICAM-1	3383	CUCCGUGAGGCCAGAGACCua	GGUCUCUGGCCUCACGGAG	-3.3	-2.1	-5.1	0	0	4	-45.6	29.3	-1.9	0	II	68.4	46.8	7	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2834	J03132	Vickers	ICAM-1	3383	AGGCACUCUCCUGCAGUGUac	ACACUGCAGGAGAGUGCCU	-2.2	-2.1	-2.1	3	0	3	-43.7	49	-0.6	3	II	57.9	44.6	0	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2835	J03132	Vickers	ICAM-1	3383	AAAGGCAGGUUGGCCAAUGag	CAUUGGCCAACCUGCCUUU	-2.1	-0.9	-3.3	2	3	4	-39.5	47	0	4	Ib	52.6	59.7	28	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2836	J03132	Vickers	ICAM-1	3383	GUAAUCUCUGAACCUGUGAac	UCACAGGUUCAGAGAUUAC	-2.4	-2.2	1.7	1	-1	2	-36.3	70.8	1	6	II	42.1	46.9	34	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2837	J03132	Vickers	ICAM-1	3383	UCCAGACAUGACCGCUGAGug	CUCAGCGGUCAUGUCUGGA	-2.1	-2.4	-1.1	-1	2	4	-42.2	45.4	2.4	2	II	57.9	45.7	33	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2838	J03132	Vickers	ICAM-1	3383	UGGAGCUGCAAUAGUGCAAgc	UUGCACUAUUGCAGCUCCA	-0.9	-2.1	-2.8	1	-1	2	-39.1	62.5	-1	5	II	47.4	49.2	5	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2839	J03132	Vickers	ICAM-1	3383	ACACAUACACACACACACAca	UGUGUGUGUGUGUAUGUGU	-2.1	-2.2	3.8	2	0	1	-36.8	55.7	-3.7	5	II	42.1	52.5	7	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2840	J03132	Vickers	ICAM-1	3383	CUGAGGUGGGAGGAUCACUug	AGUGAUCCUCCCACCUCAG	-2.1	-2.1	-0.9	1	-2	3	-43	42.7	-3.7	2	III	57.9	47.3	45	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2841	J03132	Vickers	ICAM-1	3383	GUGUGGUGUUGUGAGCCUAug	UAGGCUCACAACACCACAC	-1.3	-2.2	0.1	1	-2	3	-40.3	47.3	-0.7	3	III	52.6	35.2	41	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2842	J03132	Vickers	ICAM-1	3383	UAACACAAAGGAAGUCUGGgc	CCAGACUUCCUUUGUGUUA	-3.3	-1.3	0.4	0	4	2	-35.5	63.3	0.1	8	Ia	42.1	69.5	48	RT-PCR	T24	200nM	16h	Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents.	12500975	2003	RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.	Journal of Biological Chemistry	2003/2/1
Si2843	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCUCGCCCUUGCUCACCAU	AUGGUGAGCAAGGGCGAGG	-1.1	-3.30 	-1.10 	0.00 	-2.00 	5.00 	-44	37.90 	-8.30 	-1.00 	III	63.20 	24.10 	54.75 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 3-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2844	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCCUCGCCCUUGCUCACCA	UGGUGAGCAAGGGCGAGGA	-2.1	-2.40 	-1.10 	1.00 	1.00 	5.00 	-45.3	52.40 	1.00 	2.00 	II	63.20 	48.60 	77.71 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 4-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2845	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCCUCGCCCUUGCUCACC	GGUGAGCAAGGGCGAGGAG	-3.3	-2.10 	-1.10 	1.00 	0.00 	5.00 	-45.3	47.10 	8.40 	1.00 	II	68.40 	51.20 	60.16 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 5-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2846	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUCCUCGCCCUUGCUCAC	GUGAGCAAGGGCGAGGAGC	-2.2	-3.40 	-1.10 	1.00 	1.00 	5.00 	-45.4	42.90 	8.10 	-1.00 	II	68.40 	29.60 	38.57 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 6-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2847	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGCUCCUCGCCCUUGCUCA	UGAGCAAGGGCGAGGAGCU	-2.1	-2.10 	-2.00 	3.00 	0.00 	5.00 	-45.3	61.00 	-6.10 	3.00 	II	63.20 	44.50 	65.08 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 7-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2848	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGCUCCUCGCCCUUGCUC	GAGCAAGGGCGAGGAGCUG	-2.4	-2.10 	-2.70 	1.00 	1.00 	5.00 	-45.3	49.50 	5.40 	0.00 	II	68.40 	45.60 	44.55 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 8-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2849	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACAGCUCCUCGCCCUUGCU	AGCAAGGGCGAGGAGCUGU	-2.1	-2.20 	-2.70 	1.00 	2.00 	5.00 	-45.1	48.50 	-3.90 	1.00 	II	63.20 	44.70 	53.61 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 9-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2850	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AACAGCUCCUCGCCCUUGC	GCAAGGGCGAGGAGCUGUU	-3.4	-0.90 	-2.70 	3.00 	3.00 	5.00 	-43.9	61.40 	7.50 	4.00 	Ib	63.20 	59.30 	75.64 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 10-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2851	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAACAGCUCCUCGCCCUUG	CAAGGGCGAGGAGCUGUUC	-2.1	-2.40 	-2.70 	2.00 	1.00 	5.00 	-42.9	38.70 	0.40 	2.00 	II	63.20 	45.00 	42.30 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 11-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2852	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGAACAGCUCCUCGCCCUU	AAGGGCGAGGAGCUGUUCA	-0.9	-2.10 	-2.90 	-1.00 	1.00 	5.00 	-42.9	50.00 	-0.90 	4.00 	II	57.90 	50.60 	56.28 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 12-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2853	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGAACAGCUCCUCGCCCU	AGGGCGAGGAGCUGUUCAC	-2.1	-2.20 	-2.90 	1.00 	0.00 	5.00 	-44.2	47.70 	-6.00 	2.00 	II	63.20 	44.40 	73.97 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 13-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2854	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUGAACAGCUCCUCGCCC	GGGCGAGGAGCUGUUCACC	-3.3	-3.30 	-2.70 	1.00 	3.00 	5.00 	-45.4	35.30 	12.50 	0.00 	II	68.40 	41.10 	52.45 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 14-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2855	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGUGAACAGCUCCUCGCC	GGCGAGGAGCUGUUCACCG	-3.3	-2.40 	-2.30 	-1.00 	0.00 	4.00 	-44.5	43.20 	8.50 	0.00 	II	68.40 	43.90 	31.59 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 15-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2856	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCGGUGAACAGCUCCUCGC	GCGAGGAGCUGUUCACCGG	-3.4	-3.30 	-1.20 	0.00 	0.00 	4.00 	-44.5	45.40 	3.10 	1.00 	II	68.40 	50.40 	66.20 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 16-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2857	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCCGGUGAACAGCUCCUCG	CGAGGAGCUGUUCACCGGG	-2.4	-3.30 	-0.40 	-1.00 	0.00 	5.00 	-44.4	41.90 	5.00 	1.00 	II	68.40 	43.80 	39.28 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 17-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2858	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCCCGGUGAACAGCUCCUC	GAGGAGCUGUUCACCGGGG	-2.4	-3.30 	-0.40 	1.00 	-1.00 	6.00 	-45.3	30.30 	3.10 	1.00 	II	68.40 	40.40 	12.20 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 18-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2859	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACCCCGGUGAACAGCUCCU	AGGAGCUGUUCACCGGGGU	-2.1	-2.20 	-0.40 	2.00 	1.00 	6.00 	-45.1	39.80 	-6.60 	3.00 	II	63.20 	45.80 	50.06 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 19-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2860	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CACCCCGGUGAACAGCUCC	GGAGCUGUUCACCGGGGUG	-3.3	-2.10 	-1.70 	0.00 	0.00 	6.00 	-45.1	44.00 	10.10 	1.00 	II	68.40 	49.80 	66.02 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 20-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2861	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCACCCCGGUGAACAGCUC	GAGCUGUUCACCGGGGUGG	-2.4	-3.30 	-3.00 	1.00 	0.00 	6.00 	-45.1	28.90 	2.80 	-1.00 	II	68.40 	34.60 	51.08 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 21-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2862	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACCACCCCGGUGAACAGCU	AGCUGUUCACCGGGGUGGU	-2.1	-2.20 	-3.40 	2.00 	1.00 	6.00 	-44.9	40.50 	-6.60 	1.00 	II	63.20 	46.00 	60.60 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 22-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2863	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CACCACCCCGGUGAACAGC	GCUGUUCACCGGGGUGGUG	-3.4	-2.10 	-3.40 	1.00 	0.00 	6.00 	-44.9	38.20 	11.10 	0.00 	II	68.40 	42.60 	48.41 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 23-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2864	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCACCACCCCGGUGAACAG	CUGUUCACCGGGGUGGUGC	-2.1	-3.40 	-3.60 	1.00 	1.00 	6.00 	-44.9	30.00 	0.00 	-1.00 	II	68.40 	25.60 	33.58 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 24-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2865	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCACCACCCCGGUGAACA	UGUUCACCGGGGUGGUGCC	-2.1	-3.30 	-3.60 	1.00 	-2.00 	6.00 	-46.1	40.70 	-3.80 	0.00 	III	68.40 	30.50 	54.98 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 25-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2866	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGCACCACCCCGGUGAAC	GUUCACCGGGGUGGUGCCC	-2.2	-3.30 	-3.60 	1.00 	-1.00 	6.00 	-47.3	20.50 	5.40 	-1.00 	II	73.70 	22.90 	29.46 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 26-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2867	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGGCACCACCCCGGUGAA	UUCACCGGGGUGGUGCCCA	-0.9	-2.10 	-4.90 	-1.00 	-1.00 	6.00 	-47.2	31.50 	-6.10 	1.00 	II	68.40 	29.30 	25.44 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 27-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2868	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUGGGCACCACCCCGGUGA	UCACCGGGGUGGUGCCCAU	-2.4	-1.10 	-4.90 	2.00 	0.00 	6.00 	-47.4	44.30 	-6.10 	4.00 	II	68.40 	50.50 	44.92 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 28-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2869	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAUGGGCACCACCCCGGUG	CACCGGGGUGGUGCCCAUC	-2.1	-2.40 	-4.90 	1.00 	2.00 	6.00 	-47.4	32.70 	2.40 	0.00 	II	73.70 	38.70 	30.48 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 29-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2870	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGAUGGGCACCACCCCGGU	ACCGGGGUGGUGCCCAUCC	-2.2	-3.30 	-5.90 	0.00 	0.00 	6.00 	-48.6	24.30 	-3.90 	1.00 	III	73.70 	30.40 	28.39 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 30-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2871	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGGAUGGGCACCACCCCGG	CCGGGGUGGUGCCCAUCCU	-3.3	-2.10 	-7.80 	2.00 	3.00 	6.00 	-48.5	37.60 	-7.30 	4.00 	II	73.70 	58.20 	47.66 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 31-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2872	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGGAUGGGCACCACCCCG	CGGGGUGGUGCCCAUCCUG	-2.4	-2.10 	-9.40 	0.00 	1.00 	5.00 	-47.3	32.90 	3.10 	1.00 	II	73.70 	51.70 	32.44 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 32-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2873	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCAGGAUGGGCACCACCCC	GGGGUGGUGCCCAUCCUGG	-3.3	-3.30 	-8.20 	1.00 	1.00 	4.00 	-48.2	23.40 	5.50 	0.00 	II	73.70 	45.40 	17.03 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 33-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2874	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACCAGGAUGGGCACCACCC	GGGUGGUGCCCAUCCUGGU	-3.3	-2.20 	-7.10 	2.00 	3.00 	4.00 	-47.1	42.10 	2.50 	2.00 	II	68.40 	59.90 	61.00 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 34-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2875	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GACCAGGAUGGGCACCACC	GGUGGUGCCCAUCCUGGUC	-3.3	-2.40 	-3.40 	1.00 	1.00 	4.00 	-46.2	34.80 	12.50 	1.00 	II	68.40 	47.50 	42.56 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 35-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2876	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGACCAGGAUGGGCACCAC	GUGGUGCCCAUCCUGGUCG	-2.2	-2.40 	-1.00 	0.00 	0.00 	4.00 	-45.3	27.20 	5.40 	0.00 	II	68.40 	31.50 	27.27 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 36-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2877	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGACCAGGAUGGGCACCA	UGGUGCCCAUCCUGGUCGA	-2.1	-2.40 	-1.00 	0.00 	0.00 	4.00 	-45.5	40.40 	-3.80 	4.00 	II	63.20 	52.60 	52.46 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 37-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2878	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCGACCAGGAUGGGCACC	GGUGCCCAUCCUGGUCGAG	-3.3	-2.10 	-1.00 	-1.00 	0.00 	4.00 	-45.5	35.20 	13.40 	0.00 	II	68.40 	42.80 	30.35 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 38-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2879	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUCGACCAGGAUGGGCAC	GUGCCCAUCCUGGUCGAGC	-2.2	-3.40 	-1.00 	0.00 	1.00 	4.00 	-45.6	28.10 	7.40 	-1.00 	II	68.40 	25.20 	17.64 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 39-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2880	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGCUCGACCAGGAUGGGCA	UGCCCAUCCUGGUCGAGCU	-2.1	-2.10 	-1.00 	2.00 	1.00 	4.00 	-45.5	52.60 	-4.00 	3.00 	II	63.20 	46.70 	54.20 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 40-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2881	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGCUCGACCAGGAUGGGC	GCCCAUCCUGGUCGAGCUG	-3.4	-2.10 	-0.80 	1.00 	1.00 	4.00 	-45.5	38.60 	7.70 	2.00 	II	68.40 	46.00 	39.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 41-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2882	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCAGCUCGACCAGGAUGGG	CCCAUCCUGGUCGAGCUGG	-3.3	-3.30 	-0.80 	-2.00 	1.00 	3.00 	-45.4	23.10 	-4.40 	0.00 	II	68.40 	31.70 	33.97 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 42-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2883	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCCAGCUCGACCAGGAUGG	CCAUCCUGGUCGAGCUGGA	-3.3	-2.40 	-1.90 	1.00 	2.00 	2.00 	-44.5	52.50 	-7.30 	5.00 	II	63.20 	61.20 	64.10 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 43-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2884	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCCAGCUCGACCAGGAUG	CAUCCUGGUCGAGCUGGAC	-2.1	-2.20 	-1.90 	2.00 	0.00 	2.00 	-43.4	42.60 	5.10 	0.00 	II	63.20 	41.30 	59.28 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 44-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2885	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUCCAGCUCGACCAGGAU	AUCCUGGUCGAGCUGGACG	-1.1	-2.40 	-2.50 	0.00 	-2.00 	2.00 	-43.7	34.60 	-3.60 	0.00 	III	63.20 	24.40 	31.37 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 45-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2886	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCGUCCAGCUCGACCAGGA	UCCUGGUCGAGCUGGACGG	-2.4	-3.30 	-2.50 	0.00 	-3.00 	3.00 	-45.9	34.60 	-10.70 	0.00 	III	68.40 	29.90 	48.78 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 46-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2887	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCGUCCAGCUCGACCAGG	CCUGGUCGAGCUGGACGGC	-3.3	-3.40 	-2.50 	1.00 	0.00 	4.00 	-46.9	22.70 	-2.40 	-1.00 	II	73.70 	33.30 	33.94 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 47-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2888	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGCCGUCCAGCUCGACCAG	CUGGUCGAGCUGGACGGCG	-2.1	-2.40 	-3.30 	-1.00 	-1.00 	5.00 	-46	25.90 	1.10 	-2.00 	II	73.70 	26.00 	30.57 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 48-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2889	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGCCGUCCAGCUCGACCA	UGGUCGAGCUGGACGGCGA	-2.1	-2.40 	-3.30 	3.00 	0.00 	5.00 	-46.3	52.80 	-6.10 	3.00 	II	68.40 	56.50 	76.83 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 49-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2890	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCGCCGUCCAGCUCGACC	GGUCGAGCUGGACGGCGAC	-3.3	-2.20 	-3.30 	1.00 	2.00 	5.00 	-46.4	49.10 	14.40 	0.00 	II	73.70 	47.80 	71.54 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 50-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2891	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUCGCCGUCCAGCUCGAC	GUCGAGCUGGACGGCGACG	-2.2	-2.40 	-3.50 	0.00 	0.00 	5.00 	-45.5	24.30 	3.10 	-1.00 	II	73.70 	28.70 	24.97 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 51-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2892	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACGUCGCCGUCCAGCUCGA	UCGAGCUGGACGGCGACGU	-2.4	-2.20 	-3.00 	1.00 	0.00 	5.00 	-45.5	28.80 	-11.00 	1.00 	II	68.40 	31.40 	28.91 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 52-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2893	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UACGUCGCCGUCCAGCUCG	CGAGCUGGACGGCGACGUA	-2.4	-1.30 	-3.50 	1.00 	3.00 	5.00 	-44.4	53.70 	0.10 	5.00 	II	68.40 	70.40 	52.17 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 53-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2894	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUACGUCGCCGUCCAGCUC	GAGCUGGACGGCGACGUAA	-2.4	-0.90 	-3.50 	2.00 	4.00 	5.00 	-42.9	55.40 	12.80 	3.00 	Ib	63.20 	63.90 	59.28 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 54-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2895	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUUACGUCGCCGUCCAGCU	AGCUGGACGGCGACGUAAA	-2.1	-0.90 	-3.50 	2.00 	3.00 	5.00 	-41.4	67.90 	-6.30 	5.00 	II	57.90 	67.60 	44.73 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 55-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2896	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUUUACGUCGCCGUCCAGC	GCUGGACGGCGACGUAAAC	-3.4	-2.20 	-3.50 	1.00 	3.00 	5.00 	-41.5	53.90 	7.70 	3.00 	II	63.20 	51.50 	57.12 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 56-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2897	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUUUACGUCGCCGUCCAG	CUGGACGGCGACGUAAACG	-2.1	-2.40 	-3.50 	-2.00 	0.00 	5.00 	-40.5	42.10 	-2.00 	1.00 	II	63.20 	29.40 	37.17 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 57-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2898	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCGUUUACGUCGCCGUCCA	UGGACGGCGACGUAAACGG	-2.1	-3.30 	-3.50 	0.00 	-3.00 	5.00 	-41.7	45.50 	-5.70 	1.00 	II	63.20 	34.60 	32.38 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 58-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2899	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCGUUUACGUCGCCGUCC	GGACGGCGACGUAAACGGC	-3.3	-3.40 	-3.50 	2.00 	0.00 	5.00 	-43	43.90 	5.10 	1.00 	II	68.40 	48.70 	35.07 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 59-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2900	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCCGUUUACGUCGCCGUC	GACGGCGACGUAAACGGCC	-2.4	-3.30 	-3.50 	1.00 	0.00 	5.00 	-43	42.30 	12.80 	-1.00 	II	68.40 	34.20 	44.73 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 60-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2901	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGCCGUUUACGUCGCCGU	ACGGCGACGUAAACGGCCA	-2.2	-2.10 	-3.80 	2.00 	1.00 	5.00 	-42.7	57.00 	-6.30 	3.00 	II	63.20 	57.70 	68.59 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 61-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2902	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGGCCGUUUACGUCGCCG	CGGCGACGUAAACGGCCAC	-2.4	-2.20 	-3.80 	0.00 	3.00 	5.00 	-42.7	45.80 	2.30 	0.00 	II	68.40 	48.10 	70.94 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 62-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2903	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUGGCCGUUUACGUCGCC	GGCGACGUAAACGGCCACA	-3.3	-2.10 	-3.50 	-1.00 	4.00 	5.00 	-42.4	42.00 	9.80 	2.00 	II	63.20 	54.80 	44.03 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 63-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2904	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGUGGCCGUUUACGUCGC	GCGACGUAAACGGCCACAA	-3.4	-0.90 	-0.50 	1.00 	4.00 	5.00 	-40	58.60 	7.80 	4.00 	II	57.90 	65.80 	49.96 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 64-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2905	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUGUGGCCGUUUACGUCG	CGACGUAAACGGCCACAAG	-2.4	-2.10 	-0.50 	0.00 	1.00 	5.00 	-38.7	59.80 	1.00 	4.00 	II	57.90 	61.20 	57.87 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 10-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2906	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACUUGUGGCCGUUUACGUC	GACGUAAACGGCCACAAGU	-2.4	-2.20 	-0.50 	2.00 	4.00 	5.00 	-38.5	67.50 	12.80 	5.00 	Ib	52.60 	54.90 	53.64 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 11-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2907	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AACUUGUGGCCGUUUACGU	ACGUAAACGGCCACAAGUU	-2.2	-0.90 	-0.50 	3.00 	3.00 	5.00 	-37	71.60 	-4.00 	6.00 	II	47.40 	55.80 	53.84 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 12-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2908	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAACUUGUGGCCGUUUACG	CGUAAACGGCCACAAGUUC	-2.4	-2.40 	0.00 	2.00 	3.00 	5.00 	-37.2	59.00 	-2.40 	4.00 	II	52.60 	53.90 	68.44 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 13-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2909	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGAACUUGUGGCCGUUUAC	GUAAACGGCCACAAGUUCA	-2.2	-2.10 	0.40 	0.00 	2.00 	5.00 	-36.9	74.60 	7.50 	7.00 	Ia	47.40 	61.80 	82.92 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 14-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2910	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUGAACUUGUGGCCGUUUA	UAAACGGCCACAAGUUCAG	-1.3	-2.10 	0.40 	1.00 	-4.00 	5.00 	-36.8	64.40 	-0.70 	5.00 	II	47.40 	39.00 	55.22 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 15-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2911	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUGAACUUGUGGCCGUUU	AAACGGCCACAAGUUCAGC	-0.9	-3.40 	0.40 	0.00 	-2.00 	5.00 	-38.9	40.10 	-1.30 	3.00 	III	52.60 	30.10 	41.81 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 16-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2912	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGCUGAACUUGUGGCCGUU	AACGGCCACAAGUUCAGCG	-0.9	-2.40 	0.40 	-2.00 	-3.00 	5.00 	-40.4	37.80 	-3.00 	1.00 	III	57.90 	26.00 	42.70 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 17-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2913	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACGCUGAACUUGUGGCCGU	ACGGCCACAAGUUCAGCGU	-2.2	-2.20 	0.40 	2.00 	1.00 	5.00 	-41.7	40.40 	-4.00 	3.00 	II	57.90 	44.50 	68.07 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 18-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2914	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CACGCUGAACUUGUGGCCG	CGGCCACAAGUUCAGCGUG	-2.4	-2.10 	0.30 	-1.00 	2.00 	5.00 	-41.6	50.00 	3.40 	1.00 	II	63.20 	50.70 	63.77 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 19-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2915	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACACGCUGAACUUGUGGCC	GGCCACAAGUUCAGCGUGU	-3.3	-2.20 	-0.30 	2.00 	4.00 	4.00 	-41.4	54.90 	8.10 	4.00 	II	57.90 	56.40 	59.21 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 20-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2916	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GACACGCUGAACUUGUGGC	GCCACAAGUUCAGCGUGUC	-3.4	-2.40 	-0.30 	1.00 	3.00 	3.00 	-40.5	65.40 	9.70 	2.00 	II	57.90 	47.60 	74.71 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 21-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2917	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGACACGCUGAACUUGUGG	CCACAAGUUCAGCGUGUCC	-3.3	-3.30 	-0.30 	1.00 	2.00 	3.00 	-40.4	49.50 	9.70 	3.00 	II	57.90 	43.30 	84.19 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 22-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2918	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGACACGCUGAACUUGUG	CACAAGUUCAGCGUGUCCG	-2.1	-2.40 	0.00 	0.00 	-1.00 	3.00 	-39.5	44.00 	-4.60 	0.00 	II	57.90 	34.30 	56.35 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 23-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2919	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCGGACACGCUGAACUUGU	ACAAGUUCAGCGUGUCCGG	-2.2	-3.30 	0.70 	0.00 	-3.00 	4.00 	-40.7	37.10 	-5.90 	2.00 	III	57.90 	39.20 	53.21 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 24-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2920	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCGGACACGCUGAACUUG	CAAGUUCAGCGUGUCCGGC	-2.1	-3.40 	0.70 	1.00 	-1.00 	5.00 	-41.9	28.60 	0.70 	0.00 	II	63.20 	30.40 	57.54 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 25-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2921	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGCCGGACACGCUGAACUU	AAGUUCAGCGUGUCCGGCG	-0.9	-2.40 	-1.00 	-1.00 	-3.00 	6.00 	-42.2	31.30 	-8.30 	-1.00 	III	63.20 	28.20 	40.43 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 26-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2922	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGCCGGACACGCUGAACU	AGUUCAGCGUGUCCGGCGA	-2.1	-2.40 	-1.00 	2.00 	1.00 	6.00 	-43.7	50.70 	-1.60 	3.00 	II	63.20 	57.30 	73.34 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 27-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2923	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCGCCGGACACGCUGAAC	GUUCAGCGUGUCCGGCGAG	-2.2	-2.10 	-1.00 	-1.00 	-1.00 	6.00 	-43.7	40.40 	5.40 	0.00 	II	68.40 	38.30 	49.53 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 28-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2924	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCUCGCCGGACACGCUGAA	UUCAGCGUGUCCGGCGAGG	-0.9	-3.30 	-1.00 	0.00 	-3.00 	6.00 	-44.8	21.10 	-5.70 	0.00 	III	68.40 	16.30 	39.52 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 29-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2925	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCCUCGCCGGACACGCUGA	UCAGCGUGUCCGGCGAGGG	-2.4	-3.30 	-1.30 	0.00 	-4.00 	6.00 	-47.2	31.10 	-8.40 	0.00 	III	73.70 	25.50 	34.17 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 30-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2926	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCCUCGCCGGACACGCUG	CAGCGUGUCCGGCGAGGGC	-2.1	-3.40 	-3.90 	2.00 	0.00 	6.00 	-48.2	26.20 	2.40 	0.00 	II	78.90 	34.40 	49.31 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 31-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2927	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGCCCUCGCCGGACACGCU	AGCGUGUCCGGCGAGGGCG	-2.1	-2.40 	-5.60 	1.00 	-2.00 	6.00 	-48.5	25.40 	-8.60 	-2.00 	III	78.90 	27.90 	49.20 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 32-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2928	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGCCCUCGCCGGACACGC	GCGUGUCCGGCGAGGGCGA	-3.4	-2.40 	-5.80 	3.00 	3.00 	6.00 	-48.8	35.70 	5.10 	2.00 	II	78.90 	58.30 	38.91 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 33-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2929	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCGCCCUCGCCGGACACG	CGUGUCCGGCGAGGGCGAG	-2.4	-2.10 	-5.80 	-1.00 	0.00 	6.00 	-47.5	26.80 	-6.70 	-1.00 	II	78.90 	39.40 	39.16 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 34-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2930	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCUCGCCCUCGCCGGACAC	GUGUCCGGCGAGGGCGAGG	-2.2	-3.30 	-5.80 	-1.00 	0.00 	6.00 	-48.4	24.80 	5.50 	-1.00 	II	78.90 	27.70 	40.00 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 35-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2931	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCCUCGCCCUCGCCGGACA	UGUCCGGCGAGGGCGAGGG	-2.1	-3.30 	-5.60 	0.00 	-4.00 	6.00 	-49.5	28.20 	-5.70 	-1.00 	III	78.90 	22.30 	51.32 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 36-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2932	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCCUCGCCCUCGCCGGAC	GUCCGGCGAGGGCGAGGGC	-2.2	-3.40 	-5.60 	2.00 	0.00 	6.00 	-50.8	20.30 	5.10 	0.00 	II	84.20 	29.30 	52.80 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 37-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2933	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGCCCUCGCCCUCGCCGGA	UCCGGCGAGGGCGAGGGCG	-2.4	-2.40 	-1.40 	0.00 	-3.00 	6.00 	-51	18.70 	-2.70 	-2.00 	III	84.20 	17.50 	44.44 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 38-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2934	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGCCCUCGCCCUCGCCGG	CCGGCGAGGGCGAGGGCGA	-3.3	-2.40 	-1.20 	3.00 	3.00 	6.00 	-51	39.30 	-4.70 	2.00 	II	84.20 	55.60 	34.77 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 39-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2935	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUCGCCCUCGCCCUCGCCG	CGGCGAGGGCGAGGGCGAU	-2.4	-1.10 	-1.10 	2.00 	5.00 	5.00 	-48.8	45.40 	0.00 	1.00 	II	78.90 	56.40 	30.26 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 40-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2936	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAUCGCCCUCGCCCUCGCC	GGCGAGGGCGAGGGCGAUG	-3.3	-2.10 	-1.10 	0.00 	2.00 	5.00 	-48.5	36.80 	5.50 	0.00 	II	78.90 	43.50 	38.74 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 41-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2937	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCAUCGCCCUCGCCCUCGC	GCGAGGGCGAGGGCGAUGC	-3.4	-3.40 	-1.10 	1.00 	2.00 	5.00 	-48.6	33.20 	7.50 	0.00 	II	78.90 	35.30 	33.45 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 42-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2938	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCAUCGCCCUCGCCCUCG	CGAGGGCGAGGGCGAUGCC	-2.4	-3.30 	-2.00 	1.00 	0.00 	5.00 	-48.5	34.40 	-2.40 	2.00 	II	78.90 	41.90 	29.27 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 43-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2939	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGCAUCGCCCUCGCCCUC	GAGGGCGAGGGCGAUGCCA	-2.4	-2.10 	-3.70 	1.00 	2.00 	5.00 	-48.2	38.10 	13.20 	2.00 	II	73.70 	51.90 	30.57 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 44-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2940	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGGCAUCGCCCUCGCCCU	AGGGCGAGGGCGAUGCCAC	-2.1	-2.20 	-3.70 	2.00 	0.00 	5.00 	-48	37.90 	-8.70 	0.00 	III	73.70 	40.30 	39.74 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 45-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2941	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUGGCAUCGCCCUCGCCC	GGGCGAGGGCGAUGCCACC	-3.3	-3.30 	-3.70 	1.00 	3.00 	5.00 	-49.2	37.90 	7.40 	0.00 	II	78.90 	46.30 	16.30 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 46-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2942	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGGUGGCAUCGCCCUCGCC	GGCGAGGGCGAUGCCACCU	-3.3	-2.10 	-3.70 	1.00 	3.00 	5.00 	-48	38.10 	7.50 	1.00 	II	73.70 	48.40 	39.33 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 47-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2943	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UAGGUGGCAUCGCCCUCGC	GCGAGGGCGAUGCCACCUA	-3.4	-1.30 	-2.70 	0.00 	4.00 	5.00 	-46	49.10 	7.50 	4.00 	II	68.40 	64.80 	28.52 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 48-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2944	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUAGGUGGCAUCGCCCUCG	CGAGGGCGAUGCCACCUAC	-2.4	-2.20 	-2.70 	1.00 	2.00 	5.00 	-44.8	39.50 	-2.40 	3.00 	II	68.40 	56.90 	36.20 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 49-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2945	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUAGGUGGCAUCGCCCUC	GAGGGCGAUGCCACCUACG	-2.4	-2.40 	-2.70 	0.00 	0.00 	5.00 	-44.8	40.40 	13.10 	1.00 	II	68.40 	41.80 	32.56 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 50-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2946	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCGUAGGUGGCAUCGCCCU	AGGGCGAUGCCACCUACGG	-2.1	-3.30 	-2.70 	0.00 	-2.00 	5.00 	-45.7	33.70 	-5.60 	1.00 	III	68.40 	38.10 	24.66 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 51-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2947	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCGUAGGUGGCAUCGCCC	GGGCGAUGCCACCUACGGC	-3.3	-3.40 	-2.80 	0.00 	2.00 	5.00 	-47	32.10 	4.80 	0.00 	II	73.70 	42.20 	32.13 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 52-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2948	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCCGUAGGUGGCAUCGCC	GGCGAUGCCACCUACGGCA	-3.3	-2.10 	-4.70 	2.00 	3.00 	4.00 	-45.8	44.80 	9.80 	2.00 	II	68.40 	58.50 	48.71 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 53-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2949	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGCCGUAGGUGGCAUCGC	GCGAUGCCACCUACGGCAA	-3.4	-0.90 	-5.50 	3.00 	4.00 	4.00 	-43.4	48.60 	7.40 	3.00 	II	63.20 	67.70 	39.58 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 54-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2950	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUGCCGUAGGUGGCAUCG	CGAUGCCACCUACGGCAAG	-2.4	-2.10 	-5.50 	-2.00 	1.00 	4.00 	-42.1	52.00 	1.00 	2.00 	II	63.20 	50.00 	54.41 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 55-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2951	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUUGCCGUAGGUGGCAUC	GAUGCCACCUACGGCAAGC	-2.4	-3.40 	-5.50 	-1.00 	0.00 	4.00 	-43.1	45.50 	7.40 	1.00 	II	63.20 	33.70 	31.33 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 56-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2952	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGCUUGCCGUAGGUGGCAU	AUGCCACCUACGGCAAGCU	-1.1	-2.10 	-5.50 	2.00 	1.00 	4.00 	-42.8	49.00 	0.70 	2.00 	II	57.90 	34.30 	46.78 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 57-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2953	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGCUUGCCGUAGGUGGCA	UGCCACCUACGGCAAGCUG	-2.1	-2.10 	-5.20 	0.00 	-2.00 	4.00 	-43.8	32.00 	-5.80 	1.00 	III	63.20 	37.60 	19.27 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 58-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2954	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCAGCUUGCCGUAGGUGGC	GCCACCUACGGCAAGCUGA	-3.4	-2.40 	-4.50 	1.00 	4.00 	4.00 	-44.1	55.80 	7.80 	3.00 	Ib	63.20 	61.70 	45.26 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 59-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2955	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCAGCUUGCCGUAGGUGG	CCACCUACGGCAAGCUGAC	-3.3	-2.20 	-1.90 	2.00 	1.00 	4.00 	-42.9	54.70 	-2.40 	2.00 	II	63.20 	47.20 	55.83 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 60-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2956	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUCAGCUUGCCGUAGGUG	CACCUACGGCAAGCUGACC	-2.1	-3.30 	-1.90 	1.00 	2.00 	4.00 	-42.9	35.20 	0.30 	0.00 	II	63.20 	33.90 	34.74 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 61-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2957	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGUCAGCUUGCCGUAGGU	ACCUACGGCAAGCUGACCC	-2.2	-3.30 	-1.90 	-1.00 	-1.00 	4.00 	-44.1	37.60 	-3.90 	0.00 	III	63.20 	23.80 	39.10 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 62-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2958	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGGGUCAGCUUGCCGUAGG	CCUACGGCAAGCUGACCCU	-3.3	-2.10 	-1.90 	2.00 	2.00 	4.00 	-44	48.20 	2.40 	2.00 	II	63.20 	52.30 	67.73 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 63-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2959	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGGGUCAGCUUGCCGUAG	CUACGGCAAGCUGACCCUG	-2.1	-2.10 	-1.90 	0.00 	-1.00 	4.00 	-42.8	39.80 	1.00 	0.00 	II	63.20 	40.70 	30.78 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 64-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2960	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCAGGGUCAGCUUGCCGUA	UACGGCAAGCUGACCCUGA	-1.3	-2.40 	-1.90 	0.00 	0.00 	4.00 	-43.1	44.80 	-3.10 	3.00 	II	57.90 	48.50 	48.47 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 10-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2961	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUCAGGGUCAGCUUGCCGU	ACGGCAAGCUGACCCUGAA	-2.2	-0.90 	-1.90 	1.00 	3.00 	4.00 	-42.7	66.70 	-4.00 	5.00 	II	57.90 	65.70 	74.98 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 11-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2962	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUCAGGGUCAGCUUGCCG	CGGCAAGCUGACCCUGAAG	-2.4	-2.10 	-1.40 	0.00 	3.00 	4.00 	-42.6	48.30 	7.70 	2.00 	II	63.20 	51.90 	64.56 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 12-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2963	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACUUCAGGGUCAGCUUGCC	GGCAAGCUGACCCUGAAGU	-3.3	-2.20 	-0.20 	2.00 	4.00 	3.00 	-42.4	48.30 	5.10 	4.00 	Ib	57.90 	49.50 	62.57 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 13-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2964	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AACUUCAGGGUCAGCUUGC	GCAAGCUGACCCUGAAGUU	-3.4	-0.90 	-0.80 	1.00 	3.00 	3.00 	-40	60.40 	2.50 	7.00 	Ia	52.60 	61.40 	69.63 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 14-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2965	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAACUUCAGGGUCAGCUUG	CAAGCUGACCCUGAAGUUC	-2.1	-2.40 	-0.90 	2.00 	1.00 	3.00 	-39	58.00 	5.40 	4.00 	II	52.60 	47.20 	71.75 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 15-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2966	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGAACUUCAGGGUCAGCUU	AAGCUGACCCUGAAGUUCA	-0.9	-2.10 	-0.90 	1.00 	1.00 	3.00 	-39	72.90 	-4.00 	6.00 	II	47.40 	60.00 	91.70 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 16-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2967	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUGAACUUCAGGGUCAGCU	AGCUGACCCUGAAGUUCAU	-2.1	-1.10 	-0.60 	3.00 	3.00 	3.00 	-39.2	73.00 	1.10 	7.00 	II	47.40 	65.00 	87.31 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 17-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2968	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAUGAACUUCAGGGUCAGC	GCUGACCCUGAAGUUCAUC	-3.4	-2.40 	0.00 	-1.00 	2.00 	3.00 	-39.5	46.30 	12.40 	3.00 	II	52.60 	40.20 	75.18 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 18-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2969	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGAUGAACUUCAGGGUCAG	CUGACCCUGAAGUUCAUCU	-2.1	-2.10 	1.10 	0.00 	3.00 	3.00 	-38.2	46.60 	-2.40 	5.00 	Ia	47.40 	45.50 	69.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 19-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2970	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGAUGAACUUCAGGGUCA	UGACCCUGAAGUUCAUCUG	-2.1	-2.10 	1.10 	-1.00 	-3.00 	3.00 	-38.2	51.20 	-10.70 	5.00 	II	47.40 	45.20 	50.09 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 20-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2971	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCAGAUGAACUUCAGGGUC	GACCCUGAAGUUCAUCUGC	-2.4	-3.40 	1.20 	0.00 	1.00 	3.00 	-39.5	46.00 	15.50 	3.00 	II	52.60 	44.00 	43.50 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 21-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2972	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCAGAUGAACUUCAGGGU	ACCCUGAAGUUCAUCUGCA	-2.2	-2.10 	0.20 	1.00 	1.00 	3.00 	-39.2	67.00 	-3.30 	6.00 	II	47.40 	55.10 	78.90 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 22-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2973	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGCAGAUGAACUUCAGGG	CCCUGAAGUUCAUCUGCAC	-3.3	-2.20 	0.20 	2.00 	3.00 	3.00 	-39.2	61.10 	5.00 	4.00 	II	52.60 	56.00 	79.01 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 23-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2974	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUGCAGAUGAACUUCAGG	CCUGAAGUUCAUCUGCACC	-3.3	-3.30 	0.20 	0.00 	2.00 	2.00 	-39.2	52.20 	7.00 	2.00 	II	52.60 	41.90 	68.12 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 24-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2975	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGUGCAGAUGAACUUCAG	CUGAAGUUCAUCUGCACCA	-2.1	-2.10 	0.20 	0.00 	2.00 	2.00 	-38	62.40 	-2.60 	5.00 	II	47.40 	55.20 	67.42 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 25-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2976	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGGUGCAGAUGAACUUCA	UGAAGUUCAUCUGCACCAC	-2.1	-2.20 	0.20 	1.00 	-2.00 	2.00 	-38.1	49.00 	-6.30 	4.00 	III	47.40 	44.50 	45.42 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 26-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2977	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUGGUGCAGAUGAACUUC	GAAGUUCAUCUGCACCACC	-2.4	-3.30 	0.20 	0.00 	0.00 	2.00 	-39.3	50.80 	12.70 	3.00 	II	52.60 	43.70 	53.05 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 27-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2978	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGUGGUGCAGAUGAACUU	AAGUUCAUCUGCACCACCG	-0.9	-2.40 	0.20 	0.00 	-3.00 	3.00 	-39.3	47.30 	-6.00 	3.00 	III	52.60 	37.40 	58.69 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 28-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2979	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCGGUGGUGCAGAUGAACU	AGUUCAUCUGCACCACCGG	-2.1	-3.30 	0.20 	0.00 	-2.00 	4.00 	-41.7	44.60 	-3.90 	1.00 	III	57.90 	41.60 	38.37 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 29-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2980	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCGGUGGUGCAGAUGAAC	GUUCAUCUGCACCACCGGC	-2.2	-3.40 	0.40 	0.00 	-1.00 	5.00 	-43	28.00 	5.10 	0.00 	II	63.20 	33.00 	13.09 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 30-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2981	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCCGGUGGUGCAGAUGAA	UUCAUCUGCACCACCGGCA	-0.9	-2.10 	-0.60 	2.00 	-1.00 	5.00 	-42.9	35.40 	-8.70 	3.00 	II	57.90 	34.80 	21.77 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 31-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2982	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGCCGGUGGUGCAGAUGA	UCAUCUGCACCACCGGCAA	-2.4	-0.90 	-1.40 	2.00 	0.00 	5.00 	-42.9	50.70 	-1.40 	4.00 	II	57.90 	57.70 	48.50 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 32-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2983	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUGCCGGUGGUGCAGAUG	CAUCUGCACCACCGGCAAG	-2.1	-2.10 	-1.40 	-1.00 	0.00 	5.00 	-42.6	46.20 	1.00 	1.00 	II	63.20 	44.80 	40.68 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 33-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2984	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUUGCCGGUGGUGCAGAU	AUCUGCACCACCGGCAAGC	-1.1	-3.40 	-1.40 	0.00 	-1.00 	5.00 	-43.9	35.50 	-4.00 	0.00 	III	63.20 	16.60 	6.00 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 34-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2985	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGCUUGCCGGUGGUGCAGA	UCUGCACCACCGGCAAGCU	-2.4	-2.10 	-1.40 	2.00 	0.00 	5.00 	-44.9	40.30 	-1.40 	2.00 	II	63.20 	37.50 	4.35 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 35-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2986	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGCUUGCCGGUGGUGCAG	CUGCACCACCGGCAAGCUG	-2.1	-2.10 	-1.40 	0.00 	0.00 	5.00 	-44.6	39.80 	1.00 	1.00 	II	68.40 	39.60 	32.40 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 36-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2987	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCAGCUUGCCGGUGGUGCA	UGCACCACCGGCAAGCUGC	-2.1	-3.40 	-1.30 	1.00 	-1.00 	5.00 	-45.9	37.70 	-3.80 	0.00 	III	68.40 	29.10 	3.82 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 37-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2988	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCAGCUUGCCGGUGGUGC	GCACCACCGGCAAGCUGCC	-3.4	-3.30 	-3.20 	2.00 	1.00 	5.00 	-47.1	41.70 	7.40 	1.00 	II	73.70 	38.10 	35.37 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 38-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2989	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGCAGCUUGCCGGUGGUG	CACCACCGGCAAGCUGCCC	-2.1	-3.30 	-4.00 	0.00 	0.00 	5.00 	-47	17.60 	-2.00 	-1.00 	II	73.70 	26.10 	20.78 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 39-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2990	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGGCAGCUUGCCGGUGGU	ACCACCGGCAAGCUGCCCG	-2.2	-2.40 	-4.00 	-2.00 	-2.00 	5.00 	-47.3	24.20 	-5.90 	-1.00 	III	73.70 	23.20 	4.65 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 40-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2991	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACGGGCAGCUUGCCGGUGG	CCACCGGCAAGCUGCCCGU	-3.3	-2.20 	-4.00 	2.00 	2.00 	5.00 	-47.3	34.20 	2.40 	2.00 	II	73.70 	48.60 	38.78 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 41-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2992	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CACGGGCAGCUUGCCGGUG	CACCGGCAAGCUGCCCGUG	-2.1	-2.10 	-4.00 	0.00 	0.00 	5.00 	-46.1	24.40 	1.00 	-1.00 	II	73.70 	37.40 	19.62 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 42-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2993	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCACGGGCAGCUUGCCGGU	ACCGGCAAGCUGCCCGUGC	-2.2	-3.40 	-4.00 	0.00 	0.00 	5.00 	-47.4	20.70 	-3.30 	0.00 	III	73.70 	27.50 	16.84 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 43-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2994	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCACGGGCAGCUUGCCGG	CCGGCAAGCUGCCCGUGCC	-3.3	-3.30 	-4.00 	1.00 	2.00 	5.00 	-48.5	40.20 	0.00 	1.00 	II	78.90 	40.50 	46.16 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 44-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2995	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGCACGGGCAGCUUGCCG	CGGCAAGCUGCCCGUGCCC	-2.4	-3.30 	-4.00 	2.00 	2.00 	5.00 	-48.5	28.90 	9.70 	0.00 	II	78.90 	35.80 	27.60 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 45-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2996	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGGGCACGGGCAGCUUGCC	GGCAAGCUGCCCGUGCCCU	-3.3	-2.10 	-3.70 	2.00 	3.00 	5.00 	-48.2	27.60 	5.10 	0.00 	II	73.70 	44.60 	30.66 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 46-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2997	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGGGCACGGGCAGCUUGC	GCAAGCUGCCCGUGCCCUG	-3.4	-2.10 	-0.50 	-1.00 	0.00 	5.00 	-47	32.30 	0.50 	2.00 	II	73.70 	46.30 	56.01 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 47-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2998	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCAGGGCACGGGCAGCUUG	CAAGCUGCCCGUGCCCUGG	-2.1	-3.30 	-0.50 	0.00 	-1.00 	5.00 	-46.9	25.00 	0.40 	1.00 	II	73.70 	37.60 	46.31 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 48-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si2999	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCAGGGCACGGGCAGCUU	AAGCUGCCCGUGCCCUGGC	-0.9	-3.40 	-1.90 	0.00 	-2.00 	5.00 	-48.2	19.40 	-4.00 	0.00 	III	73.70 	22.50 	18.03 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 49-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3000	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCCAGGGCACGGGCAGCU	AGCUGCCCGUGCCCUGGCC	-2.1	-3.30 	-2.70 	1.00 	-1.00 	5.00 	-50.6	12.70 	-3.60 	1.00 	III	78.90 	27.50 	8.49 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 50-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3001	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGCCAGGGCACGGGCAGC	GCUGCCCGUGCCCUGGCCC	-3.4	-3.30 	-2.70 	1.00 	0.00 	5.00 	-51.8	11.80 	7.40 	-1.00 	II	84.20 	27.30 	28.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 51-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3002	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGGCCAGGGCACGGGCAG	CUGCCCGUGCCCUGGCCCA	-2.1	-2.10 	-2.70 	0.00 	2.00 	5.00 	-50.5	27.20 	0.10 	2.00 	II	78.90 	46.70 	26.39 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 52-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3003	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGGGCCAGGGCACGGGCA	UGCCCGUGCCCUGGCCCAC	-2.1	-2.20 	-2.70 	1.00 	-1.00 	5.00 	-50.6	19.00 	-8.70 	-1.00 	III	78.90 	29.10 	23.68 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 53-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3004	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUGGGCCAGGGCACGGGC	GCCCGUGCCCUGGCCCACC	-3.4	-3.30 	-2.70 	0.00 	2.00 	5.00 	-51.8	19.40 	9.80 	-1.00 	II	84.20 	33.00 	14.96 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 54-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3005	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGUGGGCCAGGGCACGGG	CCCGUGCCCUGGCCCACCC	-3.3	-3.30 	-1.50 	1.00 	1.00 	5.00 	-51.7	19.90 	0.00 	1.00 	II	84.20 	34.60 	8.36 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 55-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3006	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGGGUGGGCCAGGGCACGG	CCGUGCCCUGGCCCACCCU	-3.3	-2.10 	-1.50 	1.00 	3.00 	5.00 	-50.5	22.20 	2.30 	2.00 	II	78.90 	46.90 	5.09 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 56-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3007	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAGGGUGGGCCAGGGCACG	CGUGCCCUGGCCCACCCUC	-2.4	-2.40 	-1.50 	0.00 	1.00 	5.00 	-49.6	15.00 	-2.40 	0.00 	II	78.90 	38.80 	29.25 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 57-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3008	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGAGGGUGGGCCAGGGCAC	GUGCCCUGGCCCACCCUCG	-2.2	-2.40 	-1.50 	0.00 	0.00 	5.00 	-49.6	15.30 	-1.90 	0.00 	II	78.90 	36.40 	0.00 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 58-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3009	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACGAGGGUGGGCCAGGGCA	UGCCCUGGCCCACCCUCGU	-2.1	-2.20 	-1.50 	2.00 	0.00 	5.00 	-49.6	26.90 	-3.70 	2.00 	II	73.70 	40.50 	0.01 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 59-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3010	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CACGAGGGUGGGCCAGGGC	GCCCUGGCCCACCCUCGUG	-3.4	-2.10 	-1.50 	-1.00 	1.00 	5.00 	-49.6	24.40 	10.50 	1.00 	II	78.90 	41.60 	22.78 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 60-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3011	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCACGAGGGUGGGCCAGGG	CCCUGGCCCACCCUCGUGA	-3.3	-2.40 	-1.00 	2.00 	4.00 	5.00 	-48.6	26.60 	0.00 	3.00 	II	73.70 	51.80 	25.31 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 61-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3012	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCACGAGGGUGGGCCAGG	CCUGGCCCACCCUCGUGAC	-3.3	-2.20 	-1.00 	0.00 	1.00 	5.00 	-47.5	26.00 	0.00 	1.00 	II	73.70 	41.40 	30.95 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 62-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3013	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUCACGAGGGUGGGCCAG	CUGGCCCACCCUCGUGACC	-2.1	-3.30 	-1.00 	1.00 	1.00 	5.00 	-47.5	23.70 	5.70 	1.00 	II	73.70 	26.70 	24.32 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 63-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3014	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGUCACGAGGGUGGGCCA	UGGCCCACCCUCGUGACCA	-2.1	-2.10 	-1.00 	-1.00 	0.00 	5.00 	-47.5	41.00 	-3.80 	2.00 	II	68.40 	39.50 	71.33 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 64-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3015	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGGUCACGAGGGUGGGCC	GGCCCACCCUCGUGACCAC	-3.3	-2.20 	-1.00 	1.00 	3.00 	5.00 	-47.6	43.10 	9.70 	2.00 	II	73.70 	50.90 	50.96 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 10-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3016	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUGGUCACGAGGGUGGGC	GCCCACCCUCGUGACCACC	-3.4	-3.30 	-1.00 	0.00 	2.00 	4.00 	-47.6	26.40 	9.70 	-1.00 	II	73.70 	32.00 	43.92 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 11-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3017	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGUGGUCACGAGGGUGGG	CCCACCCUCGUGACCACCC	-3.3	-3.30 	-1.00 	0.00 	1.00 	3.00 	-47.5	26.70 	0.00 	0.00 	II	73.70 	31.00 	20.39 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 12-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3018	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGGGUGGUCACGAGGGUGG	CCACCCUCGUGACCACCCU	-3.3	-2.10 	-1.00 	1.00 	2.00 	3.00 	-46.3	39.00 	-4.90 	4.00 	II	68.40 	52.70 	33.94 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 13-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3019	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGGGUGGUCACGAGGGUG	CACCCUCGUGACCACCCUG	-2.1	-2.10 	-0.70 	-1.00 	0.00 	3.00 	-45.1	31.40 	-2.10 	1.00 	II	68.40 	44.20 	26.50 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 14-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3020	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCAGGGUGGUCACGAGGGU	ACCCUCGUGACCACCCUGA	-2.2	-2.40 	1.10 	1.00 	2.00 	3.00 	-45.4	29.50 	-3.90 	4.00 	II	63.20 	46.40 	37.21 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 15-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3021	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCAGGGUGGUCACGAGGG	CCCUCGUGACCACCCUGAC	-3.3	-2.20 	1.10 	1.00 	2.00 	3.00 	-45.4	35.40 	-4.90 	2.00 	II	68.40 	47.50 	42.44 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 16-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3022	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUCAGGGUGGUCACGAGG	CCUCGUGACCACCCUGACC	-3.3	-3.30 	1.10 	1.00 	1.00 	3.00 	-45.4	35.20 	8.10 	2.00 	II	68.40 	39.40 	50.73 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 17-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3023	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGGUCAGGGUGGUCACGAG	CUCGUGACCACCCUGACCU	-2.1	-2.10 	1.10 	2.00 	2.00 	3.00 	-44.2	44.10 	0.00 	3.00 	II	63.20 	39.30 	56.98 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 18-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3024	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UAGGUCAGGGUGGUCACGA	UCGUGACCACCCUGACCUA	-2.4	-1.30 	1.10 	1.00 	2.00 	3.00 	-43.4	49.80 	-1.40 	5.00 	II	57.90 	59.90 	51.78 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 19-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3025	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUAGGUCAGGGUGGUCACG	CGUGACCACCCUGACCUAC	-2.4	-2.20 	1.10 	0.00 	2.00 	3.00 	-43.2	37.30 	3.00 	1.00 	II	63.20 	43.70 	47.12 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 20-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3026	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUAGGUCAGGGUGGUCAC	GUGACCACCCUGACCUACG	-2.2	-2.40 	1.30 	-1.00 	0.00 	3.00 	-43.2	47.30 	5.40 	2.00 	II	63.20 	38.70 	66.14 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 21-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3027	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCGUAGGUCAGGGUGGUCA	UGACCACCCUGACCUACGG	-2.1	-3.30 	0.60 	0.00 	-3.00 	3.00 	-44.3	44.30 	-0.70 	4.00 	III	63.20 	36.40 	41.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 22-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3028	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCGUAGGUCAGGGUGGUC	GACCACCCUGACCUACGGC	-2.4	-3.40 	0.20 	-1.00 	0.00 	4.00 	-45.6	24.70 	9.70 	1.00 	II	68.40 	27.40 	30.04 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 23-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3029	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGCCGUAGGUCAGGGUGGU	ACCACCCUGACCUACGGCG	-2.2	-2.40 	0.20 	0.00 	-2.00 	5.00 	-45.6	20.10 	-8.30 	0.00 	III	68.40 	24.30 	45.21 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 24-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3030	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACGCCGUAGGUCAGGGUGG	CCACCCUGACCUACGGCGU	-3.3	-2.20 	0.20 	3.00 	2.00 	5.00 	-45.6	33.50 	-4.90 	3.00 	II	68.40 	50.00 	56.78 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 25-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3031	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CACGCCGUAGGUCAGGGUG	CACCCUGACCUACGGCGUG	-2.1	-2.10 	0.20 	-1.00 	0.00 	5.00 	-44.4	46.10 	3.40 	1.00 	II	68.40 	41.60 	74.33 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 26-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3032	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCACGCCGUAGGUCAGGGU	ACCCUGACCUACGGCGUGC	-2.2	-3.40 	0.20 	1.00 	0.00 	5.00 	-45.7	31.00 	-4.00 	1.00 	III	68.40 	27.40 	67.12 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 27-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3033	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCACGCCGUAGGUCAGGG	CCCUGACCUACGGCGUGCA	-3.3	-2.10 	0.20 	1.00 	4.00 	5.00 	-45.6	47.90 	4.60 	3.00 	II	68.40 	53.20 	55.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 28-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3034	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUGCACGCCGUAGGUCAGG	CCUGACCUACGGCGUGCAG	-3.3	-2.10 	0.30 	0.00 	0.00 	5.00 	-44.4	27.60 	0.70 	2.00 	II	68.40 	44.20 	47.98 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 29-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3035	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACUGCACGCCGUAGGUCAG	CUGACCUACGGCGUGCAGU	-2.1	-2.20 	0.60 	1.00 	3.00 	5.00 	-43.3	41.50 	0.40 	2.00 	II	63.20 	38.10 	56.34 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 30-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3036	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CACUGCACGCCGUAGGUCA	UGACCUACGGCGUGCAGUG	-2.1	-2.10 	0.30 	0.00 	-3.00 	5.00 	-43.3	47.10 	-8.10 	3.00 	III	63.20 	38.30 	57.96 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 31-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3037	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCACUGCACGCCGUAGGUC	GACCUACGGCGUGCAGUGC	-2.4	-3.40 	0.10 	2.00 	2.00 	5.00 	-44.6	29.30 	5.00 	1.00 	II	68.40 	36.00 	60.14 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 32-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3038	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGCACUGCACGCCGUAGGU	ACCUACGGCGUGCAGUGCU	-2.2	-2.10 	-1.50 	0.00 	1.00 	5.00 	-44.3	46.80 	-3.90 	2.00 	II	63.20 	39.40 	55.68 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 33-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3039	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AAGCACUGCACGCCGUAGG	CCUACGGCGUGCAGUGCUU	-3.3	-0.90 	-1.60 	4.00 	3.00 	5.00 	-43	54.80 	2.80 	4.00 	Ib	63.20 	59.80 	70.09 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 34-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3040	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAAGCACUGCACGCCGUAG	CUACGGCGUGCAGUGCUUC	-2.1	-2.40 	-1.60 	1.00 	1.00 	5.00 	-42.1	42.00 	-0.10 	1.00 	II	63.20 	37.00 	59.69 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 35-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3041	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGAAGCACUGCACGCCGUA	UACGGCGUGCAGUGCUUCA	-1.3	-2.10 	-1.60 	-1.00 	0.00 	5.00 	-42.1	41.60 	-3.70 	4.00 	II	57.90 	48.90 	56.49 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 36-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3042	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUGAAGCACUGCACGCCGU	ACGGCGUGCAGUGCUUCAG	-2.2	-2.10 	-1.60 	0.00 	-1.00 	5.00 	-42.9	38.70 	-8.20 	1.00 	III	63.20 	40.50 	62.10 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 37-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3043	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUGAAGCACUGCACGCCG	CGGCGUGCAGUGCUUCAGC	-2.4	-3.40 	-1.60 	0.00 	2.00 	5.00 	-44.1	29.80 	5.10 	1.00 	II	68.40 	41.40 	59.09 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 38-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3044	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCUGAAGCACUGCACGCC	GGCGUGCAGUGCUUCAGCC	-3.3	-3.30 	-2.10 	1.00 	1.00 	4.00 	-45	38.30 	8.10 	1.00 	II	68.40 	41.20 	71.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 39-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3045	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGCUGAAGCACUGCACGC	GCGUGCAGUGCUUCAGCCG	-3.4	-2.40 	-0.80 	0.00 	0.00 	4.00 	-44.1	36.00 	5.40 	1.00 	II	68.40 	43.90 	50.73 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 40-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3046	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCGGCUGAAGCACUGCACG	CGUGCAGUGCUUCAGCCGC	-2.4	-3.40 	-0.30 	0.00 	1.00 	5.00 	-44.1	37.70 	2.40 	0.00 	II	68.40 	41.90 	79.54 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 41-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3047	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGCGGCUGAAGCACUGCAC	GUGCAGUGCUUCAGCCGCU	-2.2	-2.10 	-0.60 	2.00 	2.00 	5.00 	-43.8	48.60 	2.50 	3.00 	II	63.20 	49.80 	90.24 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 42-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3048	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UAGCGGCUGAAGCACUGCA	UGCAGUGCUUCAGCCGCUA	-2.1	-1.30 	-1.30 	2.00 	1.00 	5.00 	-42.9	50.40 	1.00 	3.00 	II	57.90 	55.90 	81.99 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 43-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3049	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUAGCGGCUGAAGCACUGC	GCAGUGCUUCAGCCGCUAC	-3.4	-2.20 	-1.30 	0.00 	2.00 	5.00 	-43	39.80 	9.70 	2.00 	II	63.20 	50.40 	73.57 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 44-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3050	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUAGCGGCUGAAGCACUG	CAGUGCUUCAGCCGCUACC	-2.1	-3.30 	-1.30 	0.00 	1.00 	5.00 	-42.9	34.90 	-2.60 	1.00 	II	63.20 	36.50 	81.25 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 45-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3051	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGUAGCGGCUGAAGCACU	AGUGCUUCAGCCGCUACCC	-2.1	-3.30 	-1.00 	1.00 	-2.00 	5.00 	-44.1	29.20 	-3.90 	0.00 	III	63.20 	32.40 	87.12 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 46-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3052	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGGUAGCGGCUGAAGCAC	GUGCUUCAGCCGCUACCCC	-2.2	-3.30 	-0.30 	1.00 	0.00 	5.00 	-45.3	26.80 	8.10 	1.00 	II	68.40 	32.90 	58.04 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 47-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3053	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGGGUAGCGGCUGAAGCA	UGCUUCAGCCGCUACCCCG	-2.1	-2.40 	-0.30 	-1.00 	-3.00 	5.00 	-45.5	28.50 	-8.10 	0.00 	III	68.40 	32.20 	58.61 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 48-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3054	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGGGGUAGCGGCUGAAGC	GCUUCAGCCGCUACCCCGA	-3.4	-2.40 	-0.30 	2.00 	3.00 	5.00 	-45.8	46.30 	12.10 	3.00 	II	68.40 	60.40 	76.26 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 49-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3055	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCGGGGUAGCGGCUGAAG	CUUCAGCCGCUACCCCGAC	-2.1	-2.20 	0.90 	0.00 	0.00 	5.00 	-44.6	30.60 	-2.40 	1.00 	II	68.40 	35.10 	60.09 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 50-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3056	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUCGGGGUAGCGGCUGAA	UUCAGCCGCUACCCCGACC	-0.9	-3.30 	0.90 	0.00 	-2.00 	5.00 	-45.8	19.80 	-6.10 	2.00 	III	68.40 	17.90 	36.62 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 51-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3057	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGUCGGGGUAGCGGCUGA	UCAGCCGCUACCCCGACCA	-2.4	-2.10 	0.90 	0.00 	-1.00 	5.00 	-47	38.80 	1.00 	5.00 	II	68.40 	39.90 	58.40 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 52-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3058	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGGUCGGGGUAGCGGCUG	CAGCCGCUACCCCGACCAC	-2.1	-2.20 	0.90 	1.00 	1.00 	5.00 	-46.8	25.50 	0.00 	2.00 	II	73.70 	42.20 	37.35 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 53-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3059	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUGGUCGGGGUAGCGGCU	AGCCGCUACCCCGACCACA	-2.1	-2.10 	0.90 	0.00 	2.00 	5.00 	-46.8	38.40 	-3.50 	3.00 	II	68.40 	48.70 	33.58 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 54-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3060	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUGUGGUCGGGGUAGCGGC	GCCGCUACCCCGACCACAU	-3.4	-1.10 	0.90 	3.00 	4.00 	5.00 	-45.8	49.50 	7.40 	4.00 	Ib	68.40 	61.00 	44.70 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 55-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3061	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAUGUGGUCGGGGUAGCGG	CCGCUACCCCGACCACAUG	-3.3	-2.10 	0.90 	0.00 	3.00 	5.00 	-44.5	42.60 	0.30 	3.00 	II	68.40 	49.50 	44.37 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 56-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3062	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCAUGUGGUCGGGGUAGCG	CGCUACCCCGACCACAUGA	-2.4	-2.40 	1.20 	-1.00 	4.00 	5.00 	-43.6	44.50 	0.00 	5.00 	Ib	63.20 	58.20 	40.68 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 57-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3063	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUCAUGUGGUCGGGGUAGC	GCUACCCCGACCACAUGAA	-3.4	-0.90 	2.80 	1.00 	3.00 	5.00 	-42.1	56.00 	5.10 	6.00 	Ia	57.90 	63.00 	44.75 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 58-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3064	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUCAUGUGGUCGGGGUAG	CUACCCCGACCACAUGAAG	-2.1	-2.10 	2.80 	0.00 	0.00 	5.00 	-40.8	43.60 	-1.70 	4.00 	II	57.90 	44.00 	39.67 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 59-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3065	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUUCAUGUGGUCGGGGUA	UACCCCGACCACAUGAAGC	-1.3	-3.40 	2.80 	0.00 	-2.00 	5.00 	-42.1	46.20 	1.70 	2.00 	II	57.90 	22.30 	30.87 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 60-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3066	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCUUCAUGUGGUCGGGGU	ACCCCGACCACAUGAAGCA	-2.2	-2.10 	2.80 	1.00 	1.00 	5.00 	-42.9	65.00 	-4.00 	5.00 	II	57.90 	50.40 	62.79 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 61-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3067	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUGCUUCAUGUGGUCGGGG	CCCCGACCACAUGAAGCAG	-3.3	-2.10 	2.80 	0.00 	2.00 	5.00 	-42.8	43.80 	0.30 	1.00 	II	63.20 	48.80 	32.70 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 62-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3068	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUGCUUCAUGUGGUCGGG	CCCGACCACAUGAAGCAGC	-3.3	-3.40 	2.80 	0.00 	2.00 	4.00 	-42.9	46.10 	3.00 	1.00 	II	63.20 	35.00 	40.48 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 63-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3069	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCUGCUUCAUGUGGUCGG	CCGACCACAUGAAGCAGCA	-3.3	-2.10 	2.20 	1.00 	3.00 	3.00 	-41.7	66.30 	0.00 	5.00 	II	57.90 	62.10 	50.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 64-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3070	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGCUGCUUCAUGUGGUCG	CGACCACAUGAAGCAGCAC	-2.4	-2.20 	1.70 	1.00 	2.00 	2.00 	-40.6	58.60 	7.70 	2.00 	II	57.90 	48.30 	62.39 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 10-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3071	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUGCUGCUUCAUGUGGUC	GACCACAUGAAGCAGCACG	-2.4	-2.40 	1.70 	-2.00 	0.00 	2.00 	-40.6	47.50 	3.10 	1.00 	II	57.90 	39.10 	50.06 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 11-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3072	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGUGCUGCUUCAUGUGGU	ACCACAUGAAGCAGCACGA	-2.2	-2.40 	1.70 	2.00 	2.00 	2.00 	-40.6	61.30 	-6.60 	5.00 	II	52.60 	55.50 	70.72 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 12-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3073	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCGUGCUGCUUCAUGUGG	CCACAUGAAGCAGCACGAC	-3.3	-2.20 	1.70 	1.00 	1.00 	2.00 	-40.6	52.60 	5.40 	2.00 	II	57.90 	44.60 	67.17 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 13-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3074	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGUCGUGCUGCUUCAUGUG	CACAUGAAGCAGCACGACU	-2.1	-2.10 	1.70 	2.00 	3.00 	2.00 	-39.4	59.40 	0.70 	4.00 	II	52.60 	52.80 	64.46 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 14-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3075	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AAGUCGUGCUGCUUCAUGU	ACAUGAAGCAGCACGACUU	-2.2	-0.90 	1.60 	3.00 	2.00 	2.00 	-38.2	77.40 	-4.00 	6.00 	II	47.40 	58.10 	82.71 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 15-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3076	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAAGUCGUGCUGCUUCAUG	CAUGAAGCAGCACGACUUC	-2.1	-2.40 	-0.90 	1.00 	2.00 	2.00 	-38.4	56.80 	4.70 	4.00 	II	52.60 	48.90 	66.20 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 16-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3077	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGAAGUCGUGCUGCUUCAU	AUGAAGCAGCACGACUUCU	-1.1	-2.10 	-2.10 	1.00 	1.00 	2.00 	-38.4	55.10 	-3.30 	3.00 	II	47.40 	42.50 	67.42 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 17-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3078	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AAGAAGUCGUGCUGCUUCA	UGAAGCAGCACGACUUCUU	-2.1	-0.90 	-2.10 	2.00 	0.00 	2.00 	-38.2	62.40 	-3.40 	6.00 	II	47.40 	51.80 	63.55 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 18-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3079	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAAGAAGUCGUGCUGCUUC	GAAGCAGCACGACUUCUUC	-2.4	-2.40 	-2.10 	1.00 	2.00 	2.00 	-38.5	54.60 	14.80 	4.00 	II	52.60 	51.90 	64.69 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 19-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3080	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGAAGAAGUCGUGCUGCUU	AAGCAGCACGACUUCUUCA	-0.9	-2.10 	1.00 	0.00 	1.00 	2.00 	-38.2	57.90 	-1.00 	6.00 	II	47.40 	54.20 	50.99 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 20-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3081	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGAAGAAGUCGUGCUGCU	AGCAGCACGACUUCUUCAA	-2.1	-0.90 	1.00 	0.00 	2.00 	2.00 	-38.2	59.20 	-3.60 	7.00 	II	47.40 	61.60 	52.94 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 21-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3082	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUGAAGAAGUCGUGCUGC	GCAGCACGACUUCUUCAAG	-3.4	-2.10 	1.00 	-1.00 	2.00 	2.00 	-38.2	51.20 	8.10 	5.00 	II	52.60 	54.20 	63.47 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 22-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3083	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACUUGAAGAAGUCGUGCUG	CAGCACGACUUCUUCAAGU	-2.1	-2.20 	0.40 	0.00 	3.00 	2.00 	-37	61.20 	5.40 	6.00 	Ia	47.40 	50.20 	77.79 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 23-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3084	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GACUUGAAGAAGUCGUGCU	AGCACGACUUCUUCAAGUC	-2.1	-2.40 	-2.40 	1.00 	0.00 	2.00 	-37.3	67.30 	-1.70 	4.00 	II	47.40 	46.60 	76.48 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 24-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3085	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGACUUGAAGAAGUCGUGC	GCACGACUUCUUCAAGUCC	-3.4	-3.30 	-3.20 	1.00 	2.00 	2.00 	-38.5	53.70 	12.00 	3.00 	II	52.60 	48.60 	78.22 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 25-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3086	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGACUUGAAGAAGUCGUG	CACGACUUCUUCAAGUCCG	-2.1	-2.40 	-3.20 	-1.00 	-1.00 	3.00 	-37.5	53.50 	-4.60 	2.00 	II	52.60 	44.70 	75.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 26-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3087	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCGGACUUGAAGAAGUCGU	ACGACUUCUUCAAGUCCGC	-2.2	-3.40 	-3.20 	2.00 	-1.00 	4.00 	-38.8	49.10 	-1.60 	2.00 	III	52.60 	39.90 	52.06 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 27-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3088	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCGGACUUGAAGAAGUCG	CGACUUCUUCAAGUCCGCC	-2.4	-3.30 	-3.20 	0.00 	0.00 	5.00 	-39.9	35.80 	2.30 	-1.00 	II	57.90 	37.00 	44.01 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 28-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3089	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGCGGACUUGAAGAAGUC	GACUUCUUCAAGUCCGCCA	-2.4	-2.10 	-2.90 	0.00 	2.00 	5.00 	-39.6	43.70 	4.80 	3.00 	II	52.60 	56.90 	56.78 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 29-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3090	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUGGCGGACUUGAAGAAGU	ACUUCUUCAAGUCCGCCAU	-2.2	-1.10 	0.70 	2.00 	1.00 	5.00 	-38.3	50.90 	-6.60 	4.00 	II	47.40 	56.40 	58.93 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 30-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3091	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAUGGCGGACUUGAAGAAG	CUUCUUCAAGUCCGCCAUG	-2.1	-2.10 	1.90 	-1.00 	0.00 	5.00 	-38.2	44.40 	1.00 	3.00 	II	52.60 	42.70 	51.38 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 31-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3092	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCAUGGCGGACUUGAAGAA	UUCUUCAAGUCCGCCAUGC	-0.9	-3.40 	2.00 	0.00 	-2.00 	5.00 	-39.5	38.00 	-3.10 	3.00 	III	52.60 	20.70 	31.18 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 32-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3093	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCAUGGCGGACUUGAAGA	UCUUCAAGUCCGCCAUGCC	-2.4	-3.30 	1.20 	1.00 	-2.00 	5.00 	-41.9	50.20 	-1.50 	2.00 	III	57.90 	33.10 	43.94 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 33-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3094	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGCAUGGCGGACUUGAAG	CUUCAAGUCCGCCAUGCCC	-2.1	-3.30 	1.20 	2.00 	0.00 	5.00 	-42.8	40.90 	7.40 	1.00 	II	63.20 	34.50 	51.05 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 34-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3095	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGGCAUGGCGGACUUGAA	UUCAAGUCCGCCAUGCCCG	-0.9	-2.40 	1.20 	1.00 	-4.00 	5.00 	-43.1	33.80 	-8.30 	0.00 	III	63.20 	21.60 	37.37 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 35-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3096	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGGGCAUGGCGGACUUGA	UCAAGUCCGCCAUGCCCGA	-2.4	-2.40 	1.20 	1.00 	-1.00 	5.00 	-44.6	38.30 	-6.10 	4.00 	II	63.20 	48.30 	47.23 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 36-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3097	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUCGGGCAUGGCGGACUUG	CAAGUCCGCCAUGCCCGAA	-2.1	-0.90 	1.20 	-1.00 	2.00 	5.00 	-43.1	37.80 	-2.40 	4.00 	II	63.20 	54.70 	48.80 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 37-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3098	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUCGGGCAUGGCGGACUU	AAGUCCGCCAUGCCCGAAG	-0.9	-2.10 	1.20 	-1.00 	-1.00 	5.00 	-43.1	34.60 	-3.60 	2.00 	III	63.20 	33.70 	40.83 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 38-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3099	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCUUCGGGCAUGGCGGACU	AGUCCGCCAUGCCCGAAGG	-2.1	-3.30 	1.10 	0.00 	-2.00 	5.00 	-45.5	37.80 	-6.00 	3.00 	III	68.40 	38.90 	47.08 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 39-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3100	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCUUCGGGCAUGGCGGAC	GUCCGCCAUGCCCGAAGGC	-2.2	-3.40 	-2.10 	0.00 	0.00 	5.00 	-46.8	30.90 	12.70 	1.00 	II	73.70 	25.40 	86.60 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 40-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3101	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGCCUUCGGGCAUGGCGGA	UCCGCCAUGCCCGAAGGCU	-2.4	-2.10 	-3.30 	2.00 	0.00 	5.00 	-46.7	29.80 	-6.10 	1.00 	II	68.40 	31.40 	46.09 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 41-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3102	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UAGCCUUCGGGCAUGGCGG	CCGCCAUGCCCGAAGGCUA	-3.3	-1.30 	-3.30 	3.00 	5.00 	5.00 	-45.6	58.20 	-2.60 	3.00 	Ib	68.40 	69.10 	67.66 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 42-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3103	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUAGCCUUCGGGCAUGGCG	CGCCAUGCCCGAAGGCUAC	-2.4	-2.20 	-3.30 	2.00 	3.00 	5.00 	-44.5	50.30 	2.40 	1.00 	II	68.40 	51.50 	61.92 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 43-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3104	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUAGCCUUCGGGCAUGGC	GCCAUGCCCGAAGGCUACG	-3.4	-2.40 	-3.30 	-1.00 	1.00 	5.00 	-44.5	40.00 	5.40 	0.00 	II	68.40 	36.60 	67.60 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 44-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3105	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACGUAGCCUUCGGGCAUGG	CCAUGCCCGAAGGCUACGU	-3.3	-2.20 	-3.30 	0.00 	2.00 	5.00 	-43.3	36.70 	0.40 	3.00 	II	63.20 	44.00 	55.84 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 45-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3106	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GACGUAGCCUUCGGGCAUG	CAUGCCCGAAGGCUACGUC	-2.1	-2.40 	-3.30 	0.00 	0.00 	5.00 	-42.4	33.00 	-2.40 	1.00 	II	63.20 	37.10 	37.33 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 46-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3107	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGACGUAGCCUUCGGGCAU	AUGCCCGAAGGCUACGUCC	-1.1	-3.30 	-3.30 	1.00 	-1.00 	5.00 	-43.6	33.30 	1.50 	0.00 	III	63.20 	28.10 	17.64 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 47-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3108	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGACGUAGCCUUCGGGCA	UGCCCGAAGGCUACGUCCA	-2.1	-2.10 	-3.30 	2.00 	0.00 	5.00 	-44.6	55.40 	-3.10 	3.00 	II	63.20 	52.20 	53.02 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 48-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3109	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUGGACGUAGCCUUCGGGC	GCCCGAAGGCUACGUCCAG	-3.4	-2.10 	-3.30 	-1.00 	2.00 	5.00 	-44.6	46.50 	5.70 	1.00 	II	68.40 	51.00 	39.42 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 49-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3110	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCUGGACGUAGCCUUCGGG	CCCGAAGGCUACGUCCAGG	-3.3	-3.30 	-0.30 	-1.00 	2.00 	4.00 	-44.5	37.80 	0.40 	1.00 	II	68.40 	38.00 	44.19 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 50-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3111	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCCUGGACGUAGCCUUCGG	CCGAAGGCUACGUCCAGGA	-3.3	-2.40 	-0.30 	1.00 	3.00 	3.00 	-43.6	52.70 	4.70 	4.00 	II	63.20 	55.70 	37.91 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 51-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3112	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCCUGGACGUAGCCUUCG	CGAAGGCUACGUCCAGGAG	-2.4	-2.10 	-0.30 	1.00 	0.00 	3.00 	-42.4	42.00 	-2.00 	3.00 	II	63.20 	53.10 	43.91 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 52-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3113	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUCCUGGACGUAGCCUUC	GAAGGCUACGUCCAGGAGC	-2.4	-3.40 	-0.30 	0.00 	0.00 	3.00 	-43.4	44.10 	7.80 	2.00 	II	63.20 	36.70 	38.30 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 53-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3114	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGCUCCUGGACGUAGCCUU	AAGGCUACGUCCAGGAGCG	-0.9	-2.40 	-1.40 	1.00 	-3.00 	3.00 	-43.4	49.70 	-8.30 	1.00 	III	63.20 	35.70 	50.78 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 54-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3115	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCGCUCCUGGACGUAGCCU	AGGCUACGUCCAGGAGCGC	-2.1	-3.40 	-1.40 	2.00 	0.00 	4.00 	-45.9	33.90 	-1.70 	-1.00 	III	68.40 	30.10 	43.17 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 55-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3116	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCGCUCCUGGACGUAGCC	GGCUACGUCCAGGAGCGCA	-3.3	-2.10 	-1.40 	-1.00 	3.00 	4.00 	-45.9	46.90 	9.80 	1.00 	II	68.40 	55.50 	68.67 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 56-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3117	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGCGCUCCUGGACGUAGC	GCUACGUCCAGGAGCGCAC	-3.4	-2.20 	-1.40 	3.00 	1.00 	4.00 	-44.8	45.30 	4.80 	0.00 	II	68.40 	45.60 	57.42 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 57-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3118	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUGCGCUCCUGGACGUAG	CUACGUCCAGGAGCGCACC	-2.1	-3.30 	-0.60 	1.00 	0.00 	4.00 	-44.7	34.40 	0.00 	1.00 	II	68.40 	32.70 	45.22 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 58-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3119	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGUGCGCUCCUGGACGUA	UACGUCCAGGAGCGCACCA	-1.3	-2.10 	-0.90 	-1.00 	-1.00 	4.00 	-44.7	36.30 	-3.10 	4.00 	II	63.20 	38.50 	37.43 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 59-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3120	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUGGUGCGCUCCUGGACGU	ACGUCCAGGAGCGCACCAU	-2.2	-1.10 	-0.90 	1.00 	2.00 	4.00 	-44.5	41.20 	-8.70 	3.00 	II	63.20 	46.30 	67.79 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 60-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3121	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAUGGUGCGCUCCUGGACG	CGUCCAGGAGCGCACCAUC	-2.4	-2.40 	-0.90 	1.00 	3.00 	4.00 	-44.7	44.10 	2.40 	2.00 	II	68.40 	49.80 	54.01 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 61-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3122	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGAUGGUGCGCUCCUGGAC	GUCCAGGAGCGCACCAUCU	-2.2	-2.10 	-0.90 	3.00 	3.00 	4.00 	-44.4	52.00 	10.50 	4.00 	Ib	63.20 	51.10 	79.03 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 62-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3123	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AAGAUGGUGCGCUCCUGGA	UCCAGGAGCGCACCAUCUU	-2.4	-0.90 	-0.90 	2.00 	1.00 	4.00 	-43.1	55.10 	-6.10 	5.00 	II	57.90 	49.10 	68.25 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 63-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3124	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAAGAUGGUGCGCUCCUGG	CCAGGAGCGCACCAUCUUC	-3.3	-2.40 	-0.90 	0.00 	3.00 	4.00 	-43.1	46.10 	5.10 	4.00 	II	63.20 	51.20 	56.42 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 64-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3125	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGAAGAUGGUGCGCUCCUG	CAGGAGCGCACCAUCUUCU	-2.1	-2.10 	-1.70 	3.00 	3.00 	4.00 	-41.9	45.30 	-2.40 	3.00 	Ia	57.90 	50.90 	56.80 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 10-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3126	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AAGAAGAUGGUGCGCUCCU	AGGAGCGCACCAUCUUCUU	-2.1	-0.90 	-1.50 	1.00 	2.00 	4.00 	-40.7	50.10 	1.10 	5.00 	II	52.60 	56.90 	93.83 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 11-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3127	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAAGAAGAUGGUGCGCUCC	GGAGCGCACCAUCUUCUUC	-3.3	-2.40 	-1.00 	0.00 	2.00 	4.00 	-41	46.90 	13.10 	3.00 	II	57.90 	51.60 	60.87 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 12-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3128	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGAAGAAGAUGGUGCGCUC	GAGCGCACCAUCUUCUUCA	-2.4	-2.10 	0.90 	-1.00 	3.00 	4.00 	-39.8	54.70 	7.40 	6.00 	Ia	52.60 	58.70 	72.63 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 13-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3129	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGAAGAAGAUGGUGCGCU	AGCGCACCAUCUUCUUCAA	-2.1	-0.90 	3.40 	1.00 	3.00 	4.00 	-38.3	61.60 	1.10 	8.00 	II	47.40 	70.50 	74.80 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 14-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3130	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUGAAGAAGAUGGUGCGC	GCGCACCAUCUUCUUCAAG	-3.4	-2.10 	2.40 	-2.00 	2.00 	4.00 	-38.3	47.30 	13.40 	4.00 	II	52.60 	47.60 	49.89 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 15-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3131	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCUUGAAGAAGAUGGUGCG	CGCACCAUCUUCUUCAAGG	-2.4	-3.30 	1.10 	-2.00 	1.00 	3.00 	-38.2	52.50 	-2.00 	4.00 	II	52.60 	43.90 	70.54 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 16-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3132	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCCUUGAAGAAGAUGGUGC	GCACCAUCUUCUUCAAGGA	-3.4	-2.40 	0.20 	1.00 	3.00 	2.00 	-38.2	74.70 	9.40 	8.00 	Ib	47.40 	69.50 	79.17 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 17-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3133	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCCUUGAAGAAGAUGGUG	CACCAUCUUCUUCAAGGAC	-2.1	-2.20 	0.00 	1.00 	1.00 	2.00 	-37	52.60 	2.30 	3.00 	II	47.40 	45.90 	57.07 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 18-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3134	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUCCUUGAAGAAGAUGGU	ACCAUCUUCUUCAAGGACG	-2.2	-2.40 	0.00 	0.00 	-1.00 	2.00 	-37.3	53.60 	-8.60 	3.00 	III	47.40 	37.90 	60.43 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 19-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3135	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGUCCUUGAAGAAGAUGG	CCAUCUUCUUCAAGGACGA	-3.3	-2.40 	0.00 	2.00 	2.00 	2.00 	-37.5	77.70 	-0.30 	7.00 	II	47.40 	69.70 	82.71 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 20-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3136	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCGUCCUUGAAGAAGAUG	CAUCUUCUUCAAGGACGAC	-2.1	-2.20 	0.00 	0.00 	0.00 	2.00 	-36.4	52.60 	2.30 	2.00 	II	47.40 	43.70 	40.19 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 21-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3137	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUCGUCCUUGAAGAAGAU	AUCUUCUUCAAGGACGACG	-1.1	-2.40 	0.00 	-1.00 	-2.00 	2.00 	-36.7	40.80 	-8.60 	1.00 	III	47.40 	25.70 	37.67 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 22-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3138	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCGUCGUCCUUGAAGAAGA	UCUUCUUCAAGGACGACGG	-2.4	-3.30 	0.30 	1.00 	-4.00 	3.00 	-38.9	47.90 	-8.30 	2.00 	III	52.60 	37.30 	29.21 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 23-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3139	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCGUCGUCCUUGAAGAAG	CUUCUUCAAGGACGACGGC	-2.1	-3.40 	0.30 	1.00 	-1.00 	4.00 	-39.9	42.10 	3.00 	1.00 	II	57.90 	34.10 	31.41 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 24-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3140	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCCGUCGUCCUUGAAGAA	UUCUUCAAGGACGACGGCA	-0.9	-2.10 	0.30 	0.00 	-1.00 	4.00 	-39.9	47.70 	-3.10 	3.00 	II	52.60 	35.40 	44.72 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 25-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3141	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGCCGUCGUCCUUGAAGA	UCUUCAAGGACGACGGCAA	-2.4	-0.90 	0.50 	3.00 	1.00 	4.00 	-39.9	65.60 	-6.10 	5.00 	II	52.60 	58.70 	69.09 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 26-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3142	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUUGCCGUCGUCCUUGAAG	CUUCAAGGACGACGGCAAC	-2.1	-2.20 	1.30 	1.00 	2.00 	4.00 	-39.7	58.40 	2.40 	2.00 	II	57.90 	45.20 	57.47 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 27-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3143	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGUUGCCGUCGUCCUUGAA	UUCAAGGACGACGGCAACU	-0.9	-2.10 	1.30 	1.00 	0.00 	4.00 	-39.7	57.40 	1.70 	4.00 	II	52.60 	30.30 	53.96 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 28-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3144	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UAGUUGCCGUCGUCCUUGA	UCAAGGACGACGGCAACUA	-2.4	-1.30 	1.50 	1.00 	0.00 	4.00 	-40.1	65.10 	-6.10 	7.00 	II	52.60 	52.80 	72.14 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 29-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3145	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUAGUUGCCGUCGUCCUUG	CAAGGACGACGGCAACUAC	-2.1	-2.20 	1.60 	1.00 	2.00 	4.00 	-39.9	53.10 	0.00 	4.00 	II	57.90 	55.60 	47.07 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 30-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3146	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUAGUUGCCGUCGUCCUU	AAGGACGACGGCAACUACA	-0.9	-2.10 	1.60 	1.00 	1.00 	4.00 	-39.9	65.10 	1.50 	6.00 	II	52.60 	53.80 	63.40 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 31-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3147	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGUAGUUGCCGUCGUCCU	AGGACGACGGCAACUACAA	-2.1	-0.90 	1.90 	2.00 	2.00 	4.00 	-39.9	69.80 	-3.60 	7.00 	II	52.60 	66.00 	62.52 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 32-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3148	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUGUAGUUGCCGUCGUCC	GGACGACGGCAACUACAAG	-3.3	-2.10 	3.10 	-1.00 	2.00 	4.00 	-39.9	56.80 	3.00 	4.00 	II	57.90 	57.90 	53.23 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 33-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3149	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCUUGUAGUUGCCGUCGUC	GACGACGGCAACUACAAGA	-2.4	-2.40 	2.30 	-1.00 	3.00 	4.00 	-39	61.20 	7.50 	6.00 	Ia	52.60 	57.60 	57.85 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 34-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3150	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCUUGUAGUUGCCGUCGU	ACGACGGCAACUACAAGAC	-2.2	-2.20 	2.30 	2.00 	0.00 	4.00 	-38.8	56.90 	-1.60 	4.00 	II	52.60 	40.90 	44.83 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 35-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3151	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUCUUGUAGUUGCCGUCG	CGACGGCAACUACAAGACC	-2.4	-3.30 	1.90 	1.00 	1.00 	4.00 	-39.9	55.90 	3.00 	3.00 	II	57.90 	48.30 	80.45 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 36-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3152	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGUCUUGUAGUUGCCGUC	GACGGCAACUACAAGACCC	-2.4	-3.30 	1.10 	0.00 	0.00 	4.00 	-40.8	62.10 	10.40 	2.00 	II	57.90 	43.40 	70.03 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 37-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3153	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGGUCUUGUAGUUGCCGU	ACGGCAACUACAAGACCCG	-2.2	-2.40 	-0.20 	1.00 	-1.00 	4.00 	-40.8	57.20 	-1.30 	1.00 	III	57.90 	39.50 	69.33 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 38-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3154	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCGGGUCUUGUAGUUGCCG	CGGCAACUACAAGACCCGC	-2.4	-3.40 	-0.10 	0.00 	2.00 	5.00 	-42	36.80 	2.30 	-1.00 	II	63.20 	41.90 	49.60 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 39-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3155	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGCGGGUCUUGUAGUUGCC	GGCAACUACAAGACCCGCG	-3.3	-2.40 	0.20 	-1.00 	0.00 	6.00 	-42	42.40 	5.80 	0.00 	II	63.20 	41.00 	50.65 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 40-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3156	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCGCGGGUCUUGUAGUUGC	GCAACUACAAGACCCGCGC	-3.4	-3.40 	0.20 	2.00 	0.00 	7.00 	-42.1	40.50 	7.40 	1.00 	II	63.20 	40.60 	47.84 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 41-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3157	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCGCGGGUCUUGUAGUUG	CAACUACAAGACCCGCGCC	-2.1	-3.30 	2.30 	0.00 	0.00 	8.00 	-42	39.00 	5.40 	1.00 	II	63.20 	35.10 	53.41 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 42-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3158	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGCGCGGGUCUUGUAGUU	AACUACAAGACCCGCGCCG	-0.9	-2.40 	2.30 	0.00 	-3.00 	9.00 	-42.3	32.20 	-5.30 	0.00 	III	63.20 	24.20 	34.59 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 43-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3159	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGGCGCGGGUCUUGUAGU	ACUACAAGACCCGCGCCGA	-2.2	-2.40 	1.90 	1.00 	1.00 	9.00 	-43.8	46.70 	-4.00 	2.00 	II	63.20 	49.50 	54.04 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 44-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3160	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCGGCGCGGGUCUUGUAG	CUACAAGACCCGCGCCGAG	-2.1	-2.10 	1.90 	0.00 	0.00 	9.00 	-43.7	47.50 	5.80 	2.00 	II	68.40 	39.90 	74.80 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 45-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3161	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCUCGGCGCGGGUCUUGUA	UACAAGACCCGCGCCGAGG	-1.3	-3.30 	0.10 	1.00 	-3.00 	9.00 	-44.9	30.70 	-5.80 	0.00 	III	68.40 	21.90 	52.18 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 46-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3162	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACCUCGGCGCGGGUCUUGU	ACAAGACCCGCGCCGAGGU	-2.2	-2.20 	-0.80 	2.00 	1.00 	9.00 	-45.8	43.50 	-1.60 	4.00 	II	68.40 	39.40 	58.90 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 47-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3163	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CACCUCGGCGCGGGUCUUG	CAAGACCCGCGCCGAGGUG	-2.1	-2.10 	-1.50 	0.00 	-1.00 	9.00 	-45.7	30.10 	-4.40 	2.00 	II	73.70 	37.30 	44.87 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 48-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3164	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCACCUCGGCGCGGGUCUU	AAGACCCGCGCCGAGGUGA	-0.9	-2.40 	-2.50 	1.00 	1.00 	9.00 	-46	33.80 	-6.30 	2.00 	II	68.40 	41.00 	47.57 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 49-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3165	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUCACCUCGGCGCGGGUCU	AGACCCGCGCCGAGGUGAA	-2.1	-0.90 	-2.50 	1.00 	1.00 	9.00 	-46	54.50 	-3.90 	5.00 	II	68.40 	57.10 	79.32 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 50-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3166	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUCACCUCGGCGCGGGUC	GACCCGCGCCGAGGUGAAG	-2.4	-2.10 	-2.50 	1.00 	1.00 	9.00 	-46	38.40 	5.80 	1.00 	II	73.70 	42.40 	71.87 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 51-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3167	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACUUCACCUCGGCGCGGGU	ACCCGCGCCGAGGUGAAGU	-2.2	-2.20 	-2.50 	0.00 	2.00 	9.00 	-45.8	38.20 	-1.60 	2.00 	II	68.40 	32.90 	57.45 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 52-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3168	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AACUUCACCUCGGCGCGGG	CCCGCGCCGAGGUGAAGUU	-3.3	-0.90 	-2.50 	2.00 	4.00 	9.00 	-44.5	49.80 	-2.40 	4.00 	Ia	68.40 	54.20 	54.95 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 53-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3169	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAACUUCACCUCGGCGCGG	CCGCGCCGAGGUGAAGUUC	-3.3	-2.40 	-2.50 	1.00 	3.00 	8.00 	-43.6	32.40 	-2.40 	1.00 	II	68.40 	43.30 	28.13 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 54-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3170	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGAACUUCACCUCGGCGCG	CGCGCCGAGGUGAAGUUCG	-2.4	-2.40 	-2.50 	-1.00 	1.00 	7.00 	-42.7	48.30 	1.10 	1.00 	II	68.40 	45.70 	76.60 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 55-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3171	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGAACUUCACCUCGGCGC	GCGCCGAGGUGAAGUUCGA	-3.4	-2.40 	-2.50 	2.00 	3.00 	6.00 	-42.7	65.50 	5.40 	5.00 	Ib	63.20 	69.00 	87.44 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 56-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3172	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCGAACUUCACCUCGGCG	CGCCGAGGUGAAGUUCGAG	-2.4	-2.10 	-2.30 	-1.00 	2.00 	5.00 	-41.4	48.40 	5.40 	0.00 	II	63.20 	47.10 	66.60 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 57-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3173	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCUCGAACUUCACCUCGGC	GCCGAGGUGAAGUUCGAGG	-3.4	-3.30 	-1.70 	0.00 	1.00 	4.00 	-42.3	37.10 	5.50 	0.00 	II	63.20 	40.30 	47.35 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 58-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3174	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCCUCGAACUUCACCUCGG	CCGAGGUGAAGUUCGAGGG	-3.3	-3.30 	-1.70 	0.00 	0.00 	3.00 	-42.2	43.80 	-6.90 	0.00 	II	63.20 	44.40 	65.11 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 59-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3175	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCCUCGAACUUCACCUCG	CGAGGUGAAGUUCGAGGGC	-2.4	-3.40 	-1.40 	1.00 	0.00 	4.00 	-42.3	44.70 	5.40 	1.00 	II	63.20 	42.00 	40.11 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 60-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3176	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGCCCUCGAACUUCACCUC	GAGGUGAAGUUCGAGGGCG	-2.4	-2.40 	-1.00 	0.00 	-1.00 	5.00 	-42.3	50.60 	6.10 	-1.00 	II	63.20 	39.70 	58.14 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 61-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3177	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGCCCUCGAACUUCACCU	AGGUGAAGUUCGAGGGCGA	-2.1	-2.40 	-1.00 	3.00 	2.00 	5.00 	-42.3	68.30 	-1.70 	3.00 	II	57.90 	63.60 	76.05 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 62-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3178	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCGCCCUCGAACUUCACC	GGUGAAGUUCGAGGGCGAC	-3.3	-2.20 	-1.00 	1.00 	2.00 	5.00 	-42.4	56.30 	14.40 	-1.00 	II	63.20 	47.70 	68.14 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 63-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3179	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUCGCCCUCGAACUUCAC	GUGAAGUUCGAGGGCGACA	-2.2	-2.10 	-1.00 	1.00 	3.00 	5.00 	-41.2	53.70 	4.80 	2.00 	II	57.90 	52.80 	68.96 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 64-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3180	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGUCGCCCUCGAACUUCA	UGAAGUUCGAGGGCGACAC	-2.1	-2.20 	-1.00 	1.00 	-2.00 	5.00 	-41.2	47.90 	-8.70 	1.00 	III	57.90 	36.50 	66.49 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 10-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3181	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUGUCGCCCUCGAACUUC	GAAGUUCGAGGGCGACACC	-2.4	-3.30 	-1.00 	1.00 	0.00 	5.00 	-42.4	40.40 	5.10 	2.00 	II	63.20 	42.40 	51.84 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 11-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3182	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGUGUCGCCCUCGAACUU	AAGUUCGAGGGCGACACCC	-0.9	-3.30 	-2.30 	0.00 	-2.00 	5.00 	-43.3	36.00 	-0.90 	-1.00 	III	63.20 	24.10 	38.57 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 12-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3183	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGGGUGUCGCCCUCGAACU	AGUUCGAGGGCGACACCCU	-2.1	-2.10 	-4.10 	3.00 	0.00 	5.00 	-44.5	49.20 	-8.70 	2.00 	II	63.20 	51.60 	53.70 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 13-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3184	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGGGUGUCGCCCUCGAAC	GUUCGAGGGCGACACCCUG	-2.2	-2.10 	-4.80 	0.00 	0.00 	5.00 	-44.5	44.60 	5.40 	0.00 	II	68.40 	47.40 	58.99 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 14-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3185	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCAGGGUGUCGCCCUCGAA	UUCGAGGGCGACACCCUGG	-0.9	-3.30 	-4.80 	0.00 	-3.00 	5.00 	-45.6	27.10 	-5.70 	0.00 	III	68.40 	25.60 	26.94 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 15-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3186	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACCAGGGUGUCGCCCUCGA	UCGAGGGCGACACCCUGGU	-2.4	-2.20 	-4.80 	2.00 	0.00 	5.00 	-46.9	32.50 	-3.70 	2.00 	II	68.40 	35.90 	32.61 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 16-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3187	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CACCAGGGUGUCGCCCUCG	CGAGGGCGACACCCUGGUG	-2.4	-2.10 	-4.80 	0.00 	0.00 	5.00 	-46.6	28.20 	-1.70 	2.00 	II	73.70 	49.00 	25.29 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 17-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3188	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCACCAGGGUGUCGCCCUC	GAGGGCGACACCCUGGUGA	-2.4	-2.40 	-4.80 	1.00 	3.00 	5.00 	-46.6	38.70 	12.80 	3.00 	II	68.40 	51.40 	39.28 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 18-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3189	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUCACCAGGGUGUCGCCCU	AGGGCGACACCCUGGUGAA	-2.1	-0.90 	-4.40 	1.00 	2.00 	5.00 	-45.1	58.00 	-4.00 	4.00 	II	63.20 	63.90 	55.57 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 19-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3190	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUUCACCAGGGUGUCGCCC	GGGCGACACCCUGGUGAAC	-3.3	-2.20 	-3.30 	2.00 	4.00 	5.00 	-45.2	44.80 	15.50 	1.00 	II	68.40 	46.70 	44.30 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 20-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3191	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUUCACCAGGGUGUCGCC	GGCGACACCCUGGUGAACC	-3.3	-3.30 	-3.20 	-1.00 	2.00 	4.00 	-45.2	44.00 	7.40 	0.00 	II	68.40 	33.20 	62.00 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 21-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3192	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGUUCACCAGGGUGUCGC	GCGACACCCUGGUGAACCG	-3.4	-2.40 	-3.20 	0.00 	1.00 	3.00 	-44.3	53.20 	7.70 	2.00 	II	68.40 	47.90 	80.95 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 22-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3193	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCGGUUCACCAGGGUGUCG	CGACACCCUGGUGAACCGC	-2.4	-3.40 	-3.20 	0.00 	0.00 	4.00 	-44.3	34.70 	2.30 	0.00 	II	68.40 	34.80 	38.44 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 23-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3194	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCGGUUCACCAGGGUGUC	GACACCCUGGUGAACCGCA	-2.4	-2.10 	-3.20 	0.00 	2.00 	4.00 	-44	43.60 	5.10 	2.00 	II	63.20 	53.40 	51.35 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 24-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3195	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUGCGGUUCACCAGGGUGU	ACACCCUGGUGAACCGCAU	-2.2	-1.10 	-3.20 	3.00 	1.00 	4.00 	-42.7	49.50 	-11.30 	4.00 	II	57.90 	55.10 	75.63 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 25-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3196	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAUGCGGUUCACCAGGGUG	CACCCUGGUGAACCGCAUC	-2.1	-2.40 	-2.80 	0.00 	2.00 	4.00 	-42.9	46.30 	2.40 	1.00 	II	63.20 	44.20 	66.00 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 26-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3197	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGAUGCGGUUCACCAGGGU	ACCCUGGUGAACCGCAUCG	-2.2	-2.40 	-2.80 	-1.00 	-1.00 	4.00 	-43.2	33.10 	-5.90 	1.00 	III	63.20 	31.40 	50.81 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 27-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3198	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGAUGCGGUUCACCAGGG	CCCUGGUGAACCGCAUCGA	-3.3	-2.40 	-2.80 	1.00 	3.00 	4.00 	-43.4	44.10 	-4.90 	3.00 	II	63.20 	59.10 	82.86 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 28-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3199	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCGAUGCGGUUCACCAGG	CCUGGUGAACCGCAUCGAG	-3.3	-2.10 	-2.80 	0.00 	0.00 	4.00 	-42.2	44.10 	6.10 	1.00 	II	63.20 	52.40 	64.13 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 29-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3200	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUCGAUGCGGUUCACCAG	CUGGUGAACCGCAUCGAGC	-2.1	-3.40 	-2.80 	3.00 	1.00 	4.00 	-42.3	44.20 	3.00 	0.00 	II	63.20 	34.90 	55.62 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 30-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3201	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGCUCGAUGCGGUUCACCA	UGGUGAACCGCAUCGAGCU	-2.1	-2.10 	-2.80 	2.00 	1.00 	4.00 	-42.3	59.60 	-1.40 	2.00 	II	57.90 	46.90 	61.68 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 31-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3202	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGCUCGAUGCGGUUCACC	GGUGAACCGCAUCGAGCUG	-3.3	-2.10 	-2.80 	0.00 	1.00 	4.00 	-42.3	46.40 	5.40 	1.00 	II	63.20 	51.70 	57.74 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 32-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3203	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCAGCUCGAUGCGGUUCAC	GUGAACCGCAUCGAGCUGA	-2.2	-2.40 	-0.80 	-1.00 	3.00 	4.00 	-41.4	48.10 	5.10 	3.00 	II	57.90 	48.10 	71.25 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 33-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3204	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUCAGCUCGAUGCGGUUCA	UGAACCGCAUCGAGCUGAA	-2.1	-0.90 	-0.80 	1.00 	0.00 	4.00 	-40.1	64.10 	-1.40 	8.00 	II	52.60 	59.90 	77.65 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 34-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3205	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUCAGCUCGAUGCGGUUC	GAACCGCAUCGAGCUGAAG	-2.4	-2.10 	-0.80 	1.00 	0.00 	4.00 	-40.1	55.60 	13.40 	3.00 	II	57.90 	48.70 	69.41 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 35-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3206	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCUUCAGCUCGAUGCGGUU	AACCGCAUCGAGCUGAAGG	-0.9	-3.30 	-0.80 	-2.00 	-2.00 	4.00 	-41	43.30 	-6.00 	2.00 	III	57.90 	28.50 	52.52 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 36-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3207	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCCUUCAGCUCGAUGCGGU	ACCGCAUCGAGCUGAAGGG	-2.2	-3.30 	-0.80 	0.00 	-1.00 	4.00 	-43.4	45.20 	-8.60 	2.00 	III	63.20 	34.10 	70.09 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 37-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3208	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCCUUCAGCUCGAUGCGG	CCGCAUCGAGCUGAAGGGC	-3.3	-3.40 	-0.80 	2.00 	1.00 	4.00 	-44.6	29.50 	-2.40 	0.00 	II	68.40 	34.30 	37.51 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 38-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3209	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCCCUUCAGCUCGAUGCG	CGCAUCGAGCUGAAGGGCA	-2.4	-2.10 	-2.10 	1.00 	3.00 	4.00 	-43.4	56.90 	3.10 	2.00 	II	63.20 	57.20 	80.91 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 39-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3210	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUGCCCUUCAGCUCGAUGC	GCAUCGAGCUGAAGGGCAU	-3.4	-1.10 	-2.10 	4.00 	3.00 	4.00 	-42.1	75.60 	5.10 	4.00 	II	57.90 	68.20 	83.65 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 40-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3211	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAUGCCCUUCAGCUCGAUG	CAUCGAGCUGAAGGGCAUC	-2.1	-2.40 	-2.10 	0.00 	2.00 	4.00 	-41.1	59.80 	7.00 	0.00 	II	57.90 	42.70 	72.67 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 41-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3212	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGAUGCCCUUCAGCUCGAU	AUCGAGCUGAAGGGCAUCG	-1.1	-2.40 	-2.10 	-1.00 	-2.00 	4.00 	-41.4	36.80 	-8.30 	0.00 	III	57.90 	25.10 	46.70 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 42-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3213	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGAUGCCCUUCAGCUCGA	UCGAGCUGAAGGGCAUCGA	-2.4	-2.40 	-2.10 	0.00 	0.00 	4.00 	-42.7	43.20 	-8.70 	3.00 	II	57.90 	47.10 	52.02 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 43-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3214	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCGAUGCCCUUCAGCUCG	CGAGCUGAAGGGCAUCGAC	-2.4	-2.20 	-2.10 	1.00 	1.00 	4.00 	-42.5	50.40 	8.10 	2.00 	II	63.20 	52.40 	65.53 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 44-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3215	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGUCGAUGCCCUUCAGCUC	GAGCUGAAGGGCAUCGACU	-2.4	-2.10 	-2.10 	4.00 	3.00 	4.00 	-42.2	54.90 	8.10 	2.00 	Ib	57.90 	52.60 	60.50 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 45-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3216	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AAGUCGAUGCCCUUCAGCU	AGCUGAAGGGCAUCGACUU	-2.1	-0.90 	-2.10 	2.00 	3.00 	4.00 	-40.7	67.10 	-6.30 	4.00 	II	52.60 	59.20 	80.01 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 46-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3217	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAAGUCGAUGCCCUUCAGC	GCUGAAGGGCAUCGACUUC	-3.4	-2.40 	-2.20 	0.00 	3.00 	4.00 	-41	53.40 	7.40 	2.00 	II	57.90 	52.00 	66.55 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 47-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3218	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGAAGUCGAUGCCCUUCAG	CUGAAGGGCAUCGACUUCA	-2.1	-2.10 	-3.90 	0.00 	3.00 	4.00 	-39.7	61.50 	0.10 	4.00 	Ib	52.60 	53.80 	74.36 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 48-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3219	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGAAGUCGAUGCCCUUCA	UGAAGGGCAUCGACUUCAA	-2.1	-0.90 	-4.60 	2.00 	0.00 	4.00 	-38.5	70.20 	1.40 	8.00 	II	47.40 	68.90 	64.04 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 49-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3220	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUGAAGUCGAUGCCCUUC	GAAGGGCAUCGACUUCAAG	-2.4	-2.10 	-3.90 	0.00 	0.00 	4.00 	-38.5	58.70 	13.40 	4.00 	II	52.60 	53.80 	59.47 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 50-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3221	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCUUGAAGUCGAUGCCCUU	AAGGGCAUCGACUUCAAGG	-0.9	-3.30 	-0.40 	-2.00 	-2.00 	4.00 	-39.4	43.10 	-6.00 	3.00 	II	52.60 	32.70 	43.18 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 51-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3222	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCCUUGAAGUCGAUGCCCU	AGGGCAUCGACUUCAAGGA	-2.1	-2.40 	-0.40 	1.00 	2.00 	4.00 	-40.9	58.50 	-6.60 	6.00 	II	52.60 	57.80 	79.95 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 52-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3223	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCCUUGAAGUCGAUGCCC	GGGCAUCGACUUCAAGGAG	-3.3	-2.10 	-0.20 	0.00 	1.00 	4.00 	-40.9	48.00 	3.10 	2.00 	II	57.90 	55.70 	58.23 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 53-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3224	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCUCCUUGAAGUCGAUGCC	GGCAUCGACUUCAAGGAGG	-3.3	-3.30 	-1.30 	0.00 	1.00 	3.00 	-40.9	58.10 	10.80 	1.00 	II	57.90 	46.00 	83.01 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 54-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3225	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCCUCCUUGAAGUCGAUGC	GCAUCGACUUCAAGGAGGA	-3.4	-2.40 	-1.30 	2.00 	2.00 	2.00 	-40	83.40 	9.70 	6.00 	II	52.60 	67.50 	91.90 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 55-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3226	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCCUCCUUGAAGUCGAUG	CAUCGACUUCAAGGAGGAC	-2.1	-2.20 	-1.30 	1.00 	1.00 	2.00 	-38.8	56.20 	4.60 	1.00 	II	52.60 	43.70 	51.73 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 56-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3227	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUCCUCCUUGAAGUCGAU	AUCGACUUCAAGGAGGACG	-1.1	-2.40 	-1.30 	-1.00 	-2.00 	2.00 	-39.1	43.40 	-8.60 	0.00 	III	52.60 	24.30 	19.56 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 57-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3228	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCGUCCUCCUUGAAGUCGA	UCGACUUCAAGGAGGACGG	-2.4	-3.30 	-1.30 	1.00 	-3.00 	3.00 	-41.3	47.20 	-8.30 	0.00 	III	57.90 	34.60 	44.20 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 58-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3229	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCGUCCUCCUUGAAGUCG	CGACUUCAAGGAGGACGGC	-2.4	-3.40 	-1.30 	1.00 	0.00 	4.00 	-42.3	43.10 	3.00 	0.00 	II	63.20 	38.70 	42.17 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 59-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3230	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCCGUCCUCCUUGAAGUC	GACUUCAAGGAGGACGGCA	-2.4	-2.10 	-1.30 	0.00 	2.00 	4.00 	-42	53.10 	8.10 	1.00 	II	57.90 	51.80 	53.68 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 60-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3231	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGCCGUCCUCCUUGAAGU	ACUUCAAGGAGGACGGCAA	-2.2	-0.90 	-1.30 	3.00 	2.00 	4.00 	-40.5	70.90 	-6.30 	4.00 	II	52.60 	64.20 	86.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 61-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3232	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUUGCCGUCCUCCUUGAAG	CUUCAAGGAGGACGGCAAC	-2.1	-2.20 	-0.80 	1.00 	2.00 	4.00 	-40.5	58.70 	2.40 	2.00 	II	57.90 	44.10 	62.85 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 62-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3233	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUUGCCGUCCUCCUUGAA	UUCAAGGAGGACGGCAACA	-0.9	-2.10 	1.30 	0.00 	0.00 	4.00 	-40.5	60.40 	-0.70 	5.00 	II	52.60 	38.30 	68.20 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 63-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3234	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUGUUGCCGUCCUCCUUGA	UCAAGGAGGACGGCAACAU	-2.4	-1.10 	1.60 	2.00 	0.00 	4.00 	-40.7	61.20 	-8.50 	6.00 	II	52.60 	47.00 	90.74 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 64-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3235	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAUGUUGCCGUCCUCCUUG	CAAGGAGGACGGCAACAUC	-2.1	-2.40 	3.10 	1.00 	2.00 	4.00 	-40.7	60.10 	2.40 	4.00 	II	57.90 	53.90 	71.47 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 10-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3236	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGAUGUUGCCGUCCUCCUU	AAGGAGGACGGCAACAUCC	-0.9	-3.30 	-0.40 	2.00 	-1.00 	4.00 	-41.9	55.20 	1.50 	2.00 	II	57.90 	36.50 	30.98 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 11-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3237	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGGAUGUUGCCGUCCUCCU	AGGAGGACGGCAACAUCCU	-2.1	-2.10 	-2.20 	3.00 	1.00 	4.00 	-43.1	58.90 	-6.30 	3.00 	II	57.90 	53.70 	68.57 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 12-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3238	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGGAUGUUGCCGUCCUCC	GGAGGACGGCAACAUCCUG	-3.3	-2.10 	-2.90 	-1.00 	1.00 	4.00 	-43.1	49.50 	5.70 	2.00 	II	63.20 	57.60 	40.72 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 13-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3239	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCAGGAUGUUGCCGUCCUC	GAGGACGGCAACAUCCUGG	-2.4	-3.30 	-2.90 	-1.00 	0.00 	4.00 	-43.1	33.70 	5.50 	0.00 	II	63.20 	38.10 	15.23 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 14-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3240	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCCAGGAUGUUGCCGUCCU	AGGACGGCAACAUCCUGGG	-2.1	-3.30 	-2.50 	0.00 	-2.00 	4.00 	-44	38.70 	-3.60 	0.00 	III	63.20 	36.80 	80.61 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 15-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3241	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCCCAGGAUGUUGCCGUCC	GGACGGCAACAUCCUGGGG	-3.3	-3.30 	-1.40 	0.00 	-1.00 	4.00 	-45.2	33.10 	11.10 	1.00 	II	68.40 	45.00 	81.66 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 16-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3242	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCCCAGGAUGUUGCCGUC	GACGGCAACAUCCUGGGGC	-2.4	-3.40 	-0.10 	0.00 	0.00 	5.00 	-45.3	33.80 	10.40 	0.00 	II	68.40 	26.50 	28.24 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 17-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3243	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCCCCAGGAUGUUGCCGU	ACGGCAACAUCCUGGGGCA	-2.2	-2.10 	-0.10 	2.00 	2.00 	5.00 	-45	55.00 	-1.60 	4.00 	II	63.20 	55.00 	84.16 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 18-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3244	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGCCCCAGGAUGUUGCCG	CGGCAACAUCCUGGGGCAC	-2.4	-2.20 	0.10 	2.00 	3.00 	5.00 	-45	45.20 	7.70 	0.00 	II	68.40 	42.90 	55.35 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 19-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3245	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUGCCCCAGGAUGUUGCC	GGCAACAUCCUGGGGCACA	-3.3	-2.10 	0.10 	-1.00 	4.00 	5.00 	-44.7	55.60 	7.40 	3.00 	II	63.20 	55.40 	68.58 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 20-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3246	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGUGCCCCAGGAUGUUGC	GCAACAUCCUGGGGCACAA	-3.4	-0.90 	0.10 	1.00 	4.00 	5.00 	-42.3	72.20 	7.10 	6.00 	II	57.90 	69.40 	84.43 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 21-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3247	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUGUGCCCCAGGAUGUUG	CAACAUCCUGGGGCACAAG	-2.1	-2.10 	0.70 	0.00 	0.00 	5.00 	-41	50.90 	0.30 	3.00 	II	57.90 	47.10 	76.97 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 22-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3248	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUUGUGCCCCAGGAUGUU	AACAUCCUGGGGCACAAGC	-0.9	-3.40 	0.70 	0.00 	-1.00 	5.00 	-42.3	34.20 	-6.30 	2.00 	III	57.90 	25.70 	39.47 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 23-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3249	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGCUUGUGCCCCAGGAUGU	ACAUCCUGGGGCACAAGCU	-2.2	-2.10 	-1.30 	2.00 	0.00 	5.00 	-43.5	51.00 	-11.30 	4.00 	II	57.90 	45.90 	55.51 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 24-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3250	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGCUUGUGCCCCAGGAUG	CAUCCUGGGGCACAAGCUG	-2.1	-2.10 	-2.00 	1.00 	-1.00 	5.00 	-43.4	45.90 	-4.30 	1.00 	II	63.20 	47.30 	45.12 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 25-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3251	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCAGCUUGUGCCCCAGGAU	AUCCUGGGGCACAAGCUGG	-1.1	-3.30 	-2.00 	0.00 	-2.00 	5.00 	-44.6	38.60 	-8.30 	0.00 	III	63.20 	32.90 	32.90 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 26-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3252	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCCAGCUUGUGCCCCAGGA	UCCUGGGGCACAAGCUGGA	-2.4	-2.40 	-2.00 	2.00 	0.00 	5.00 	-45.9	51.70 	-3.70 	3.00 	II	63.20 	47.20 	64.64 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 27-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3253	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCCAGCUUGUGCCCCAGG	CCUGGGGCACAAGCUGGAG	-3.3	-2.10 	-2.00 	0.00 	0.00 	5.00 	-45.6	36.90 	3.10 	0.00 	II	68.40 	45.50 	26.71 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 28-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3254	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACUCCAGCUUGUGCCCCAG	CUGGGGCACAAGCUGGAGU	-2.1	-2.20 	-2.00 	2.00 	3.00 	5.00 	-44.5	44.30 	3.00 	2.00 	II	63.20 	43.30 	90.87 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 29-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3255	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UACUCCAGCUUGUGCCCCA	UGGGGCACAAGCUGGAGUA	-2.1	-1.30 	-2.00 	0.00 	1.00 	5.00 	-43.7	58.00 	-3.80 	4.00 	II	57.90 	55.20 	64.89 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 30-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3256	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUACUCCAGCUUGUGCCCC	GGGGCACAAGCUGGAGUAC	-3.3	-2.20 	-2.00 	2.00 	4.00 	5.00 	-43.8	47.70 	12.80 	1.00 	II	63.20 	50.50 	50.80 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 31-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3257	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUACUCCAGCUUGUGCCC	GGGCACAAGCUGGAGUACA	-3.3	-2.10 	-2.00 	-1.00 	4.00 	4.00 	-42.6	68.80 	8.10 	3.00 	Ib	57.90 	62.20 	88.84 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 32-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3258	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGUACUCCAGCUUGUGCC	GGCACAAGCUGGAGUACAA	-3.3	-0.90 	-1.60 	2.00 	5.00 	3.00 	-40.2	86.50 	10.10 	7.00 	Ia	52.60 	79.00 	89.95 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 33-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3259	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUUGUACUCCAGCUUGUGC	GCACAAGCUGGAGUACAAC	-3.4	-2.20 	-0.50 	1.00 	3.00 	2.00 	-39.1	67.50 	14.40 	4.00 	II	52.60 	48.30 	85.91 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 34-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3260	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGUUGUACUCCAGCUUGUG	CACAAGCUGGAGUACAACU	-2.1	-2.10 	0.70 	1.00 	3.00 	2.00 	-37.8	61.30 	-2.40 	5.00 	Ia	47.40 	54.10 	73.47 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 35-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3261	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UAGUUGUACUCCAGCUUGU	ACAAGCUGGAGUACAACUA	-2.2	-1.30 	0.70 	1.00 	1.00 	2.00 	-37	72.00 	-11.30 	8.00 	II	42.10 	63.30 	77.66 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 36-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3262	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUAGUUGUACUCCAGCUUG	CAAGCUGGAGUACAACUAC	-2.1	-2.20 	0.70 	0.00 	1.00 	2.00 	-37	61.70 	0.10 	5.00 	II	47.40 	55.00 	77.60 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 37-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3263	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUAGUUGUACUCCAGCUU	AAGCUGGAGUACAACUACA	-0.9	-2.10 	0.00 	1.00 	1.00 	2.00 	-37	76.20 	-0.90 	7.00 	II	42.10 	62.10 	74.17 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 38-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3264	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGUAGUUGUACUCCAGCU	AGCUGGAGUACAACUACAA	-2.1	-0.90 	0.00 	2.00 	2.00 	2.00 	-37	79.90 	-1.30 	7.00 	II	42.10 	69.90 	81.59 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 39-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3265	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUUGUAGUUGUACUCCAGC	GCUGGAGUACAACUACAAC	-3.4	-2.20 	0.00 	0.00 	3.00 	2.00 	-37.1	61.70 	12.10 	4.00 	II	47.40 	56.30 	69.99 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 40-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3266	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUUGUAGUUGUACUCCAG	CUGGAGUACAACUACAACA	-2.1	-2.10 	0.00 	0.00 	3.00 	2.00 	-35.8	76.30 	0.40 	6.00 	Ia	42.10 	61.80 	76.29 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 41-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3267	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUGUUGUAGUUGUACUCCA	UGGAGUACAACUACAACAG	-2.1	-2.10 	0.00 	1.00 	-2.00 	2.00 	-35.8	64.60 	-5.80 	4.00 	II	42.10 	49.50 	56.26 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 42-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3268	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUGUUGUAGUUGUACUCC	GGAGUACAACUACAACAGC	-3.3	-3.40 	0.00 	0.00 	2.00 	2.00 	-37.1	64.90 	12.80 	4.00 	II	47.40 	52.30 	74.02 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 43-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3269	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCUGUUGUAGUUGUACUC	GAGUACAACUACAACAGCC	-2.4	-3.30 	0.00 	0.00 	0.00 	3.00 	-37.1	69.30 	10.40 	3.00 	II	47.40 	44.60 	85.32 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 44-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3270	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGCUGUUGUAGUUGUACU	AGUACAACUACAACAGCCA	-2.1	-2.10 	0.00 	3.00 	1.00 	3.00 	-36.8	79.10 	0.70 	5.00 	II	42.10 	61.10 	81.16 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 45-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3271	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGGCUGUUGUAGUUGUAC	GUACAACUACAACAGCCAC	-2.2	-2.20 	0.20 	0.00 	1.00 	3.00 	-36.9	61.20 	9.70 	3.00 	II	47.40 	49.20 	75.82 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 46-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3272	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUGGCUGUUGUAGUUGUA	UACAACUACAACAGCCACA	-1.3	-2.10 	2.30 	0.00 	0.00 	3.00 	-36.8	62.90 	-3.30 	5.00 	II	42.10 	45.90 	69.22 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 47-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3273	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGUGGCUGUUGUAGUUGU	ACAACUACAACAGCCACAA	-2.2	-0.90 	2.30 	1.00 	1.00 	3.00 	-36.4	70.80 	-4.00 	6.00 	II	42.10 	59.80 	70.00 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 48-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3274	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUUGUGGCUGUUGUAGUUG	CAACUACAACAGCCACAAC	-2.1	-2.20 	2.30 	0.00 	2.00 	3.00 	-36.4	60.50 	5.40 	5.00 	II	47.40 	51.80 	68.17 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 49-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3275	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUUGUGGCUGUUGUAGUU	AACUACAACAGCCACAACG	-0.9	-2.40 	2.30 	-1.00 	-2.00 	3.00 	-36.7	60.20 	-3.00 	3.00 	III	47.40 	33.00 	68.52 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 50-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3276	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACGUUGUGGCUGUUGUAGU	ACUACAACAGCCACAACGU	-2.2	-2.20 	2.20 	3.00 	1.00 	3.00 	-38	65.70 	-1.60 	5.00 	II	47.40 	50.50 	91.37 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 51-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3277	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GACGUUGUGGCUGUUGUAG	CUACAACAGCCACAACGUC	-2.1	-2.40 	2.10 	1.00 	1.00 	3.00 	-38.2	61.40 	3.00 	4.00 	II	52.60 	43.10 	85.53 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 52-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3278	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGACGUUGUGGCUGUUGUA	UACAACAGCCACAACGUCU	-1.3	-2.10 	2.10 	2.00 	0.00 	3.00 	-38.2	54.00 	-6.10 	4.00 	II	47.40 	41.70 	79.31 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 53-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3279	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UAGACGUUGUGGCUGUUGU	ACAACAGCCACAACGUCUA	-2.2	-1.30 	2.10 	2.00 	2.00 	3.00 	-38.2	79.40 	0.80 	7.00 	II	47.40 	62.20 	91.91 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 54-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3280	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUAGACGUUGUGGCUGUUG	CAACAGCCACAACGUCUAU	-2.1	-1.10 	2.10 	1.00 	3.00 	3.00 	-37.1	58.60 	2.70 	7.00 	Ib	47.40 	62.80 	78.33 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 55-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3281	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UAUAGACGUUGUGGCUGUU	AACAGCCACAACGUCUAUA	-0.9	-1.30 	2.10 	-1.00 	2.00 	3.00 	-36.3	61.60 	-1.00 	6.00 	II	42.10 	49.50 	73.12 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 56-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3282	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUAUAGACGUUGUGGCUGU	ACAGCCACAACGUCUAUAU	-2.2	-1.10 	2.10 	1.00 	2.00 	3.00 	-36.5	62.30 	-1.30 	8.00 	II	42.10 	56.50 	74.34 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 57-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3283	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAUAUAGACGUUGUGGCUG	CAGCCACAACGUCUAUAUC	-2.1	-2.40 	2.10 	1.00 	3.00 	3.00 	-36.7	62.60 	5.40 	5.00 	II	47.40 	51.10 	71.99 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 58-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3284	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGAUAUAGACGUUGUGGCU	AGCCACAACGUCUAUAUCA	-2.1	-2.10 	2.10 	-1.00 	2.00 	3.00 	-36.7	73.20 	1.80 	7.00 	II	42.10 	58.40 	75.87 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 59-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3285	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUGAUAUAGACGUUGUGGC	GCCACAACGUCUAUAUCAU	-3.4	-1.10 	2.40 	3.00 	5.00 	3.00 	-35.7	80.70 	7.40 	7.00 	Ia	42.10 	69.20 	96.68 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 60-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3286	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAUGAUAUAGACGUUGUGG	CCACAACGUCUAUAUCAUG	-3.3	-2.10 	2.50 	-1.00 	2.00 	2.00 	-34.4	73.00 	3.00 	5.00 	II	42.10 	57.90 	87.94 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 61-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3287	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCAUGAUAUAGACGUUGUG	CACAACGUCUAUAUCAUGG	-2.1	-3.30 	1.20 	-1.00 	0.00 	2.00 	-34.4	57.30 	0.40 	4.00 	II	42.10 	43.40 	64.59 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 62-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3288	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCAUGAUAUAGACGUUGU	ACAACGUCUAUAUCAUGGC	-2.2	-3.40 	-0.70 	0.00 	-2.00 	3.00 	-35.7	63.80 	-4.30 	4.00 	III	42.10 	37.50 	27.48 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 63-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3289	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCCAUGAUAUAGACGUUG	CAACGUCUAUAUCAUGGCC	-2.1	-3.30 	-1.50 	1.00 	-1.00 	4.00 	-36.8	47.30 	2.70 	3.00 	II	47.40 	45.00 	53.01 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 64-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3290	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGCCAUGAUAUAGACGUU	AACGUCUAUAUCAUGGCCG	-0.9	-2.40 	-1.50 	0.00 	-3.00 	5.00 	-37.1	46.00 	-3.20 	0.00 	III	47.40 	28.90 	33.60 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 10-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3291	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGGCCAUGAUAUAGACGU	ACGUCUAUAUCAUGGCCGA	-2.2	-2.40 	-1.50 	1.00 	1.00 	5.00 	-38.6	62.90 	-4.00 	4.00 	II	47.40 	62.80 	68.70 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 11-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3292	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCGGCCAUGAUAUAGACG	CGUCUAUAUCAUGGCCGAC	-2.4	-2.20 	-1.50 	0.00 	2.00 	5.00 	-38.6	54.90 	5.10 	0.00 	II	52.60 	47.50 	55.55 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 12-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3293	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUCGGCCAUGAUAUAGAC	GUCUAUAUCAUGGCCGACA	-2.2	-2.10 	-1.20 	1.00 	3.00 	5.00 	-38.3	60.60 	7.40 	3.00 	II	47.40 	53.90 	43.75 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 13-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3294	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGUCGGCCAUGAUAUAGA	UCUAUAUCAUGGCCGACAA	-2.4	-0.90 	0.50 	1.00 	1.00 	5.00 	-37	71.70 	-4.00 	7.00 	II	42.10 	64.10 	69.70 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 14-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3295	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUGUCGGCCAUGAUAUAG	CUAUAUCAUGGCCGACAAG	-2.1	-2.10 	0.50 	0.00 	0.00 	5.00 	-36.7	64.90 	3.30 	5.00 	II	47.40 	49.70 	60.50 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 15-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3296	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUUGUCGGCCAUGAUAUA	UAUAUCAUGGCCGACAAGC	-1.3	-3.40 	0.50 	0.00 	-3.00 	5.00 	-38	43.80 	-6.10 	2.00 	III	47.40 	20.90 	34.07 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 16-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3297	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCUUGUCGGCCAUGAUAU	AUAUCAUGGCCGACAAGCA	-1.1	-2.10 	-0.70 	2.00 	0.00 	5.00 	-38.8	70.90 	-8.90 	7.00 	II	47.40 	55.80 	85.53 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 17-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3298	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUGCUUGUCGGCCAUGAUA	UAUCAUGGCCGACAAGCAG	-1.3	-2.10 	-0.70 	1.00 	-4.00 	5.00 	-39.8	54.10 	-5.70 	3.00 	III	52.60 	41.00 	61.07 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 18-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3299	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCUGCUUGUCGGCCAUGAU	AUCAUGGCCGACAAGCAGA	-1.1	-2.40 	-0.70 	1.00 	1.00 	5.00 	-40.9	62.10 	-1.60 	5.00 	II	52.60 	48.50 	66.07 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 19-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3300	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUCUGCUUGUCGGCCAUGA	UCAUGGCCGACAAGCAGAA	-2.4	-0.90 	-0.70 	2.00 	0.00 	5.00 	-40.7	63.90 	-6.10 	7.00 	II	52.60 	55.70 	72.90 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 20-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3301	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUCUGCUUGUCGGCCAUG	CAUGGCCGACAAGCAGAAG	-2.1	-2.10 	-0.70 	-1.00 	0.00 	5.00 	-40.4	42.20 	-4.40 	2.00 	II	57.90 	46.00 	38.02 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 21-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3302	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCUUCUGCUUGUCGGCCAU	AUGGCCGACAAGCAGAAGA	-1.1	-2.40 	-0.70 	-1.00 	1.00 	5.00 	-40.7	64.40 	1.50 	6.00 	II	52.60 	46.00 	62.46 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 22-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3303	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUCUUCUGCUUGUCGGCCA	UGGCCGACAAGCAGAAGAA	-2.1	-0.90 	-0.70 	2.00 	1.00 	5.00 	-40.5	72.40 	-3.80 	7.00 	II	52.60 	61.50 	82.16 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 23-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3304	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUUCUUCUGCUUGUCGGCC	GGCCGACAAGCAGAAGAAC	-3.3	-2.20 	-0.70 	2.00 	4.00 	5.00 	-40.6	59.90 	12.80 	2.00 	II	57.90 	52.20 	63.25 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 24-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3305	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUUCUUCUGCUUGUCGGC	GCCGACAAGCAGAAGAACG	-3.4	-2.40 	-0.70 	-1.00 	1.00 	4.00 	-39.7	64.90 	6.10 	2.00 	II	57.90 	46.70 	90.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 25-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3306	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCGUUCUUCUGCUUGUCGG	CCGACAAGCAGAAGAACGG	-3.3	-3.30 	0.00 	1.00 	1.00 	3.00 	-39.6	69.00 	-2.00 	2.00 	II	57.90 	47.00 	83.89 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 26-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3307	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCGUUCUUCUGCUUGUCG	CGACAAGCAGAAGAACGGC	-2.4	-3.40 	1.40 	0.00 	1.00 	4.00 	-39.7	56.50 	4.70 	1.00 	II	57.90 	42.20 	76.05 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 27-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3308	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCCGUUCUUCUGCUUGUC	GACAAGCAGAAGAACGGCA	-2.4	-2.10 	-0.30 	2.00 	2.00 	4.00 	-39.4	64.50 	8.10 	4.00 	II	52.60 	58.10 	84.26 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 28-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3309	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUGCCGUUCUUCUGCUUGU	ACAAGCAGAAGAACGGCAU	-2.2	-1.10 	-0.30 	3.00 	1.00 	4.00 	-38.1	65.90 	-6.30 	5.00 	II	47.40 	54.90 	72.88 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 29-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3310	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAUGCCGUUCUUCUGCUUG	CAAGCAGAAGAACGGCAUC	-2.1	-2.40 	-0.30 	0.00 	2.00 	4.00 	-38.3	67.30 	7.80 	4.00 	II	52.60 	48.80 	72.26 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 30-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3311	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGAUGCCGUUCUUCUGCUU	AAGCAGAAGAACGGCAUCA	-0.9	-2.10 	-0.30 	0.00 	1.00 	4.00 	-38.3	66.50 	-3.30 	5.00 	II	47.40 	49.90 	78.94 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 31-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3312	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGAUGCCGUUCUUCUGCU	AGCAGAAGAACGGCAUCAA	-2.1	-0.90 	-0.30 	1.00 	3.00 	4.00 	-38.3	71.80 	-4.00 	6.00 	II	47.40 	65.20 	73.19 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 32-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3313	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUGAUGCCGUUCUUCUGC	GCAGAAGAACGGCAUCAAG	-3.4	-2.10 	-0.10 	0.00 	2.00 	4.00 	-38.3	67.90 	15.90 	5.00 	II	52.60 	59.80 	62.15 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 33-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3314	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCUUGAUGCCGUUCUUCUG	CAGAAGAACGGCAUCAAGG	-2.1	-3.30 	0.80 	1.00 	0.00 	4.00 	-38.2	64.70 	1.00 	3.00 	II	52.60 	42.80 	73.29 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 34-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3315	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACCUUGAUGCCGUUCUUCU	AGAAGAACGGCAUCAAGGU	-2.1	-2.20 	0.80 	2.00 	1.00 	4.00 	-38.3	71.50 	-4.00 	6.00 	II	47.40 	52.80 	82.46 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 35-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3316	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CACCUUGAUGCCGUUCUUC	GAAGAACGGCAUCAAGGUG	-2.4	-2.10 	0.20 	0.00 	0.00 	4.00 	-38.3	58.50 	3.00 	4.00 	II	52.60 	51.20 	68.43 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 36-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3317	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCACCUUGAUGCCGUUCUU	AAGAACGGCAUCAAGGUGA	-0.9	-2.40 	-0.90 	1.00 	1.00 	4.00 	-38.3	62.50 	-3.90 	5.00 	II	47.40 	50.30 	70.62 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 37-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3318	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUCACCUUGAUGCCGUUCU	AGAACGGCAUCAAGGUGAA	-2.1	-0.90 	-0.90 	2.00 	1.00 	4.00 	-38.3	82.90 	-1.60 	8.00 	II	47.40 	69.30 	91.47 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 38-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3319	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUUCACCUUGAUGCCGUUC	GAACGGCAUCAAGGUGAAC	-2.4	-2.20 	-0.90 	1.00 	1.00 	4.00 	-38.4	61.30 	15.40 	4.00 	II	52.60 	46.90 	74.28 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 39-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3320	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGUUCACCUUGAUGCCGUU	AACGGCAUCAAGGUGAACU	-0.9	-2.10 	-0.90 	0.00 	1.00 	4.00 	-38.1	56.70 	-4.00 	4.00 	II	47.40 	37.10 	59.25 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 40-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3321	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AAGUUCACCUUGAUGCCGU	ACGGCAUCAAGGUGAACUU	-2.2	-0.90 	-0.90 	2.00 	3.00 	4.00 	-38.1	67.70 	-4.20 	6.00 	II	47.40 	58.90 	87.02 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 41-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3322	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAAGUUCACCUUGAUGCCG	CGGCAUCAAGGUGAACUUC	-2.4	-2.40 	-0.90 	1.00 	3.00 	4.00 	-38.3	54.00 	3.00 	3.00 	II	52.60 	53.40 	70.62 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 42-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3323	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGAAGUUCACCUUGAUGCC	GGCAUCAAGGUGAACUUCA	-3.3	-2.10 	-0.90 	0.00 	4.00 	3.00 	-38	72.80 	8.10 	6.00 	Ia	47.40 	68.20 	94.06 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 43-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3324	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGAAGUUCACCUUGAUGC	GCAUCAAGGUGAACUUCAA	-3.4	-0.90 	-0.90 	2.00 	4.00 	2.00 	-35.6	91.40 	7.70 	9.00 	Ia	42.10 	83.90 	90.63 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 44-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3325	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUGAAGUUCACCUUGAUG	CAUCAAGGUGAACUUCAAG	-2.1	-2.10 	-0.70 	-1.00 	1.00 	2.00 	-34.3	67.60 	5.40 	4.00 	II	42.10 	51.90 	85.50 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 45-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3326	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCUUGAAGUUCACCUUGAU	AUCAAGGUGAACUUCAAGA	-1.1	-2.40 	-0.40 	0.00 	1.00 	2.00 	-34.6	66.40 	-3.90 	6.00 	II	36.80 	47.80 	77.14 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 46-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3327	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUCUUGAAGUUCACCUUGA	UCAAGGUGAACUUCAAGAU	-2.4	-1.10 	-0.40 	2.00 	0.00 	2.00 	-34.6	69.10 	-8.70 	7.00 	II	36.80 	55.40 	96.03 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 47-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3328	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAUCUUGAAGUUCACCUUG	CAAGGUGAACUUCAAGAUC	-2.1	-2.40 	-0.40 	1.00 	1.00 	2.00 	-34.6	68.40 	5.40 	5.00 	II	42.10 	54.10 	93.71 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 48-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3329	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGAUCUUGAAGUUCACCUU	AAGGUGAACUUCAAGAUCC	-0.9	-3.30 	-0.40 	1.00 	-1.00 	2.00 	-35.8	74.40 	-1.00 	4.00 	II	42.10 	42.50 	93.86 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 49-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3330	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGAUCUUGAAGUUCACCU	AGGUGAACUUCAAGAUCCG	-2.1	-2.40 	-0.10 	1.00 	-1.00 	3.00 	-37.3	75.10 	-1.30 	4.00 	III	47.40 	52.40 	93.27 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 50-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3331	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCGGAUCUUGAAGUUCACC	GGUGAACUUCAAGAUCCGC	-3.3	-3.40 	0.30 	0.00 	1.00 	4.00 	-38.6	51.00 	14.70 	1.00 	II	52.60 	45.50 	72.60 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 51-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3332	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCGGAUCUUGAAGUUCAC	GUGAACUUCAAGAUCCGCC	-2.2	-3.30 	0.50 	0.00 	0.00 	5.00 	-38.6	41.70 	4.80 	1.00 	II	52.60 	31.50 	56.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 52-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3333	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGCGGAUCUUGAAGUUCA	UGAACUUCAAGAUCCGCCA	-2.1	-2.10 	1.10 	2.00 	-1.00 	5.00 	-38.5	55.00 	-6.30 	5.00 	II	47.40 	56.70 	78.09 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 53-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3334	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGGCGGAUCUUGAAGUUC	GAACUUCAAGAUCCGCCAC	-2.4	-2.20 	1.80 	0.00 	0.00 	5.00 	-38.6	46.70 	10.40 	3.00 	II	52.60 	47.80 	75.05 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 54-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3335	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUGGCGGAUCUUGAAGUU	AACUUCAAGAUCCGCCACA	-0.9	-2.10 	2.30 	-1.00 	1.00 	5.00 	-38.3	55.30 	-3.30 	5.00 	II	47.40 	46.40 	65.81 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 55-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3336	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGUGGCGGAUCUUGAAGU	ACUUCAAGAUCCGCCACAA	-2.2	-0.90 	2.30 	1.00 	2.00 	5.00 	-38.3	68.40 	-4.00 	6.00 	II	47.40 	64.20 	84.53 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 56-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3337	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUUGUGGCGGAUCUUGAAG	CUUCAAGAUCCGCCACAAC	-2.1	-2.20 	2.80 	1.00 	2.00 	5.00 	-38.3	62.00 	10.10 	4.00 	II	52.60 	45.90 	69.02 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 57-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3338	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUUGUGGCGGAUCUUGAA	UUCAAGAUCCGCCACAACA	-0.9	-2.10 	4.80 	1.00 	0.00 	5.00 	-38.3	65.70 	-3.80 	6.00 	II	47.40 	47.00 	81.89 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 58-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3339	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUGUUGUGGCGGAUCUUGA	UCAAGAUCCGCCACAACAU	-2.4	-1.10 	5.40 	3.00 	1.00 	5.00 	-38.5	67.80 	-4.00 	7.00 	II	47.40 	54.20 	80.01 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 59-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3340	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAUGUUGUGGCGGAUCUUG	CAAGAUCCGCCACAACAUC	-2.1	-2.40 	2.70 	1.00 	1.00 	5.00 	-38.5	54.70 	-2.40 	5.00 	II	52.60 	49.00 	57.62 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 60-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3341	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGAUGUUGUGGCGGAUCUU	AAGAUCCGCCACAACAUCG	-0.9	-2.40 	1.00 	-1.00 	-2.00 	5.00 	-38.8	44.30 	-8.30 	3.00 	II	52.60 	35.50 	62.21 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 61-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3342	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGAUGUUGUGGCGGAUCU	AGAUCCGCCACAACAUCGA	-2.1	-2.40 	1.00 	1.00 	0.00 	5.00 	-40.3	62.60 	-1.60 	7.00 	II	52.60 	56.10 	80.25 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 62-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3343	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCGAUGUUGUGGCGGAUC	GAUCCGCCACAACAUCGAG	-2.4	-2.10 	1.20 	-1.00 	-1.00 	5.00 	-40.3	46.80 	8.10 	2.00 	II	57.90 	50.60 	57.92 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 63-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3344	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCUCGAUGUUGUGGCGGAU	AUCCGCCACAACAUCGAGG	-1.1	-3.30 	1.20 	0.00 	-2.00 	5.00 	-41.2	34.50 	-3.00 	1.00 	III	57.90 	22.80 	36.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 64-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3345	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCCUCGAUGUUGUGGCGGA	UCCGCCACAACAUCGAGGA	-2.4	-2.40 	-0.60 	0.00 	0.00 	5.00 	-42.5	49.20 	-3.80 	4.00 	II	57.90 	43.70 	77.77 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 10-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3346	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCCUCGAUGUUGUGGCGG	CCGCCACAACAUCGAGGAC	-3.3	-2.20 	0.20 	1.00 	3.00 	5.00 	-42.3	47.10 	5.40 	2.00 	II	63.20 	46.50 	29.22 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 11-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3347	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUCCUCGAUGUUGUGGCG	CGCCACAACAUCGAGGACG	-2.4	-2.40 	0.40 	-1.00 	1.00 	4.00 	-41.4	50.10 	1.00 	0.00 	II	63.20 	35.10 	36.41 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 12-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3348	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCGUCCUCGAUGUUGUGGC	GCCACAACAUCGAGGACGG	-3.4	-3.30 	0.40 	1.00 	1.00 	3.00 	-42.3	60.70 	7.70 	2.00 	II	63.20 	48.00 	73.44 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 13-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3349	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCGUCCUCGAUGUUGUGG	CCACAACAUCGAGGACGGC	-3.3	-3.40 	0.30 	1.00 	1.00 	4.00 	-42.3	53.10 	7.70 	1.00 	II	63.20 	36.60 	50.45 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 14-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3350	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCCGUCCUCGAUGUUGUG	CACAACAUCGAGGACGGCA	-2.1	-2.10 	0.30 	0.00 	2.00 	4.00 	-41.1	51.50 	0.00 	2.00 	II	57.90 	48.40 	42.56 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 15-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3351	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUGCCGUCCUCGAUGUUGU	ACAACAUCGAGGACGGCAG	-2.2	-2.10 	0.30 	2.00 	-1.00 	4.00 	-41.1	55.20 	-8.60 	2.00 	III	57.90 	44.40 	49.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 16-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3352	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUGCCGUCCUCGAUGUUG	CAACAUCGAGGACGGCAGC	-2.1	-3.40 	0.30 	1.00 	0.00 	4.00 	-42.3	42.50 	-2.40 	2.00 	II	63.20 	38.60 	41.07 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 17-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3353	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGCUGCCGUCCUCGAUGUU	AACAUCGAGGACGGCAGCG	-0.9	-2.40 	0.30 	-2.00 	-3.00 	4.00 	-42.6	35.30 	-2.90 	0.00 	III	63.20 	19.50 	22.07 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 18-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3354	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACGCUGCCGUCCUCGAUGU	ACAUCGAGGACGGCAGCGU	-2.2	-2.20 	-0.30 	3.00 	0.00 	4.00 	-43.9	43.80 	-8.70 	3.00 	II	63.20 	41.00 	46.68 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 19-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3355	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CACGCUGCCGUCCUCGAUG	CAUCGAGGACGGCAGCGUG	-2.1	-2.10 	-1.00 	0.00 	0.00 	4.00 	-43.8	47.20 	0.40 	0.00 	II	68.40 	49.00 	55.56 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 20-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3356	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCACGCUGCCGUCCUCGAU	AUCGAGGACGGCAGCGUGC	-1.1	-3.40 	-1.00 	3.00 	-1.00 	4.00 	-45.1	37.10 	1.50 	0.00 	III	68.40 	25.50 	29.90 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 21-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3357	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCACGCUGCCGUCCUCGA	UCGAGGACGGCAGCGUGCA	-2.4	-2.10 	-1.30 	1.00 	0.00 	4.00 	-46.1	46.70 	-6.10 	2.00 	II	68.40 	43.20 	37.45 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 22-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3358	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUGCACGCUGCCGUCCUCG	CGAGGACGGCAGCGUGCAG	-2.4	-2.10 	-1.30 	0.00 	1.00 	4.00 	-45.8	36.90 	-1.70 	2.00 	II	73.70 	51.30 	27.54 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 23-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3359	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUGCACGCUGCCGUCCUC	GAGGACGGCAGCGUGCAGC	-2.4	-3.40 	-1.30 	0.00 	1.00 	4.00 	-46.8	28.00 	7.50 	-1.00 	II	73.70 	26.50 	20.11 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 24-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3360	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGCUGCACGCUGCCGUCCU	AGGACGGCAGCGUGCAGCU	-2.1	-2.10 	-1.90 	2.00 	1.00 	4.00 	-46.5	42.80 	-1.60 	2.00 	II	68.40 	44.60 	26.23 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 25-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3361	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAGCUGCACGCUGCCGUCC	GGACGGCAGCGUGCAGCUC	-3.3	-2.40 	-2.00 	2.00 	1.00 	4.00 	-46.8	37.80 	8.10 	1.00 	II	73.70 	45.80 	39.97 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 26-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3362	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGAGCUGCACGCUGCCGUC	GACGGCAGCGUGCAGCUCG	-2.4	-2.40 	-2.00 	-2.00 	0.00 	4.00 	-45.9	34.70 	3.10 	0.00 	II	73.70 	39.50 	30.47 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 27-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3363	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCGAGCUGCACGCUGCCGU	ACGGCAGCGUGCAGCUCGC	-2.2	-3.40 	-2.00 	2.00 	0.00 	4.00 	-46.9	37.60 	-1.60 	1.00 	III	73.70 	35.30 	31.53 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 28-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3364	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCGAGCUGCACGCUGCCG	CGGCAGCGUGCAGCUCGCC	-2.4	-3.30 	-2.40 	1.00 	1.00 	4.00 	-48	23.40 	2.70 	-1.00 	II	78.90 	33.00 	27.87 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 29-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3365	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGCGAGCUGCACGCUGCC	GGCAGCGUGCAGCUCGCCG	-3.3	-2.40 	-3.20 	-1.00 	0.00 	5.00 	-48	16.70 	5.50 	-1.00 	II	78.90 	36.50 	15.60 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 30-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3366	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGGCGAGCUGCACGCUGC	GCAGCGUGCAGCUCGCCGA	-3.4	-2.40 	-2.40 	1.00 	2.00 	5.00 	-47.1	42.00 	2.50 	3.00 	II	73.70 	59.90 	45.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 31-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3367	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCGGCGAGCUGCACGCUG	CAGCGUGCAGCUCGCCGAC	-2.1	-2.20 	-2.40 	1.00 	1.00 	5.00 	-45.9	25.70 	2.40 	0.00 	II	73.70 	40.30 	30.52 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 32-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3368	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUCGGCGAGCUGCACGCU	AGCGUGCAGCUCGCCGACC	-2.1	-3.30 	-2.40 	1.00 	0.00 	5.00 	-47.1	20.70 	-3.30 	-1.00 	III	73.70 	25.60 	36.25 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 33-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3369	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGUCGGCGAGCUGCACGC	GCGUGCAGCUCGCCGACCA	-3.4	-2.10 	-2.20 	0.00 	3.00 	5.00 	-47.1	47.40 	5.10 	4.00 	II	73.70 	60.40 	62.15 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 34-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3370	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGGUCGGCGAGCUGCACG	CGUGCAGCUCGCCGACCAC	-2.4	-2.20 	-0.10 	1.00 	2.00 	5.00 	-45.9	42.90 	7.00 	2.00 	II	73.70 	48.80 	54.46 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 35-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3371	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGUGGUCGGCGAGCUGCAC	GUGCAGCUCGCCGACCACU	-2.2	-2.10 	-0.10 	2.00 	3.00 	5.00 	-45.6	34.60 	7.40 	2.00 	II	68.40 	42.10 	56.92 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 36-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3372	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UAGUGGUCGGCGAGCUGCA	UGCAGCUCGCCGACCACUA	-2.1	-1.30 	-0.10 	1.00 	1.00 	5.00 	-44.7	43.50 	-8.70 	5.00 	II	63.20 	54.90 	56.41 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 37-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3373	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUAGUGGUCGGCGAGCUGC	GCAGCUCGCCGACCACUAC	-3.4	-2.20 	-0.10 	1.00 	2.00 	5.00 	-44.8	40.60 	5.10 	4.00 	II	68.40 	52.50 	51.22 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 38-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3374	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUAGUGGUCGGCGAGCUG	CAGCUCGCCGACCACUACC	-2.1	-3.30 	2.90 	-1.00 	1.00 	5.00 	-44.7	32.70 	2.40 	2.00 	II	68.40 	33.00 	49.93 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 39-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3375	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGUAGUGGUCGGCGAGCU	AGCUCGCCGACCACUACCA	-2.1	-2.10 	3.20 	2.00 	1.00 	5.00 	-44.7	44.80 	-3.60 	5.00 	II	63.20 	52.70 	64.64 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 40-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3376	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUGGUAGUGGUCGGCGAGC	GCUCGCCGACCACUACCAG	-3.4	-2.10 	3.20 	-1.00 	0.00 	5.00 	-44.7	30.00 	3.10 	2.00 	II	68.40 	46.70 	34.38 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 41-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3377	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUGGUAGUGGUCGGCGAG	CUCGCCGACCACUACCAGC	-2.1	-3.40 	2.00 	-1.00 	1.00 	5.00 	-44.7	34.10 	5.40 	1.00 	II	68.40 	31.50 	32.31 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 42-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3378	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCUGGUAGUGGUCGGCGA	UCGCCGACCACUACCAGCA	-2.4	-2.10 	0.30 	2.00 	0.00 	5.00 	-44.7	50.00 	-3.80 	4.00 	II	63.20 	45.50 	74.12 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 43-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3379	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUGCUGGUAGUGGUCGGCG	CGCCGACCACUACCAGCAG	-2.4	-2.10 	0.30 	0.00 	2.00 	5.00 	-44.4	39.20 	0.30 	2.00 	II	68.40 	52.00 	38.65 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 44-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3380	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCUGCUGGUAGUGGUCGGC	GCCGACCACUACCAGCAGA	-3.4	-2.40 	0.60 	-1.00 	4.00 	4.00 	-44.4	52.60 	10.40 	5.00 	II	63.20 	59.10 	58.69 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 45-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3381	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUCUGCUGGUAGUGGUCGG	CCGACCACUACCAGCAGAA	-3.3	-0.90 	1.40 	1.00 	4.00 	3.00 	-41.9	63.90 	2.30 	5.00 	Ib	57.90 	60.90 	78.21 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 46-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3382	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUUCUGCUGGUAGUGGUCG	CGACCACUACCAGCAGAAC	-2.4	-2.20 	1.40 	2.00 	3.00 	2.00 	-40.8	51.00 	2.30 	3.00 	II	57.90 	50.20 	57.45 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 47-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3383	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUUCUGCUGGUAGUGGUC	GACCACUACCAGCAGAACA	-2.4	-2.10 	1.40 	-1.00 	3.00 	2.00 	-40.5	68.50 	7.80 	6.00 	Ib	52.60 	60.10 	77.82 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 48-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3384	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGUUCUGCUGGUAGUGGU	ACCACUACCAGCAGAACAC	-2.2	-2.20 	1.40 	2.00 	0.00 	2.00 	-40.3	64.30 	-4.00 	4.00 	II	52.60 	41.20 	79.21 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 49-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3385	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUGUUCUGCUGGUAGUGG	CCACUACCAGCAGAACACC	-3.3	-3.30 	1.40 	1.00 	2.00 	2.00 	-41.4	50.00 	2.30 	2.00 	II	57.90 	40.30 	52.61 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 50-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3386	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGUGUUCUGCUGGUAGUG	CACUACCAGCAGAACACCC	-2.1	-3.30 	1.40 	0.00 	0.00 	3.00 	-41.4	47.70 	0.70 	1.00 	II	57.90 	37.90 	46.68 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 51-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3387	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGGUGUUCUGCUGGUAGU	ACUACCAGCAGAACACCCC	-2.2	-3.30 	1.40 	1.00 	-2.00 	4.00 	-42.6	49.10 	-6.30 	1.00 	III	57.90 	32.20 	37.60 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 52-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3388	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGGGUGUUCUGCUGGUAG	CUACCAGCAGAACACCCCC	-2.1	-3.30 	1.90 	0.00 	0.00 	5.00 	-43.7	44.10 	4.70 	1.00 	II	63.20 	36.70 	53.07 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 53-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3389	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGGGGUGUUCUGCUGGUA	UACCAGCAGAACACCCCCA	-1.3	-2.10 		1.00 	-1.00 	5.00 	-43.7	43.40 	-3.10 	3.00 	II	57.90 	44.20 	64.58 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 54-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3390	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUGGGGGUGUUCUGCUGGU	ACCAGCAGAACACCCCCAU	-2.2	-1.10 		1.00 	2.00 	5.00 	-43.5	40.80 	-6.30 	3.00 	II	57.90 	48.00 	55.93 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 55-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3391	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAUGGGGGUGUUCUGCUGG	CCAGCAGAACACCCCCAUC	-3.3	-2.40 	3.50 	0.00 	3.00 	5.00 	-43.7	47.10 	7.80 	3.00 	II	63.20 	48.80 	37.82 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 56-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3392	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGAUGGGGGUGUUCUGCUG	CAGCAGAACACCCCCAUCG	-2.1	-2.40 	1.80 	0.00 	0.00 	5.00 	-42.8	45.00 	1.00 	2.00 	II	63.20 	39.40 	30.49 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 57-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3393	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCGAUGGGGGUGUUCUGCU	AGCAGAACACCCCCAUCGG	-2.1	-3.30 	0.70 	0.00 	-1.00 	5.00 	-44	42.10 	-3.60 	2.00 	III	63.20 	40.80 	28.80 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 58-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3394	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCGAUGGGGGUGUUCUGC	GCAGAACACCCCCAUCGGC	-3.4	-3.40 	0.30 	1.00 	1.00 	5.00 	-45.3	39.90 	15.50 	2.00 	II	68.40 	38.40 	29.42 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 59-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3395	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGCCGAUGGGGGUGUUCUG	CAGAACACCCCCAUCGGCG	-2.1	-2.40 	0.30 	1.00 	-1.00 	5.00 	-44.3	32.60 	-2.00 	0.00 	II	68.40 	31.30 	53.10 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 60-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3396	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGCCGAUGGGGGUGUUCU	AGAACACCCCCAUCGGCGA	-2.1	-2.40 	0.30 	2.00 	1.00 	5.00 	-44.6	51.30 	-1.60 	4.00 	II	63.20 	54.10 	56.63 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 61-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3397	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCGCCGAUGGGGGUGUUC	GAACACCCCCAUCGGCGAC	-2.4	-2.20 	0.30 	-1.00 	0.00 	5.00 	-44.7	36.20 	7.40 	2.00 	II	68.40 	36.50 	33.82 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 62-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3398	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUCGCCGAUGGGGGUGUU	AACACCCCCAUCGGCGACG	-0.9	-2.40 	-0.80 	-1.00 	-2.00 	5.00 	-44.7	23.70 	-6.00 	0.00 	III	68.40 	16.90 	18.55 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 63-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3399	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCGUCGCCGAUGGGGGUGU	ACACCCCCAUCGGCGACGG	-2.2	-3.30 	-2.10 	-1.00 	-3.00 	5.00 	-47.1	25.20 	-6.00 	1.00 	III	73.70 	26.70 	20.81 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 64-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3400	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCGUCGCCGAUGGGGGUG	CACCCCCAUCGGCGACGGC	-2.1	-3.40 	-2.50 	0.00 	0.00 	5.00 	-48.3	24.80 	5.30 	0.00 	II	78.90 	28.50 	19.37 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 10-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3401	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCCGUCGCCGAUGGGGGU	ACCCCCAUCGGCGACGGCC	-2.2	-3.30 	-3.00 	1.00 	-1.00 	5.00 	-49.5	19.30 	-4.00 	-2.00 	III	78.90 	18.30 	36.26 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 11-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3402	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGCCGUCGCCGAUGGGGG	CCCCCAUCGGCGACGGCCC	-3.3	-3.30 	-3.80 	3.00 	2.00 	6.00 	-50.6	24.70 	-2.60 	-1.00 	II	84.20 	34.90 	17.46 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 12-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3403	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGGCCGUCGCCGAUGGGG	CCCCAUCGGCGACGGCCCC	-3.3	-3.30 	-3.80 	1.00 	1.00 	7.00 	-50.6	20.70 	-4.70 	-1.00 	II	84.20 	33.60 	12.82 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 13-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3404	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGGGCCGUCGCCGAUGGG	CCCAUCGGCGACGGCCCCG	-3.3	-2.40 	-3.80 	-2.00 	0.00 	8.00 	-49.7	13.70 	-2.00 	-2.00 	II	84.20 	25.80 	3.78 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 14-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3405	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACGGGGCCGUCGCCGAUGG	CCAUCGGCGACGGCCCCGU	-3.3	-2.20 	-3.80 	2.00 	2.00 	8.00 	-48.6	25.30 	0.10 	1.00 	II	78.90 	42.40 	10.21 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 15-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3406	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CACGGGGCCGUCGCCGAUG	CAUCGGCGACGGCCCCGUG	-2.1	-2.10 	-3.80 	0.00 	-1.00 	8.00 	-47.4	21.50 	-4.40 	0.00 	II	78.90 	43.30 	33.38 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 16-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3407	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCACGGGGCCGUCGCCGAU	AUCGGCGACGGCCCCGUGC	-1.1	-3.40 	-3.80 	1.00 	-1.00 	8.00 	-48.7	13.60 	1.50 	0.00 	III	78.90 	19.20 	35.75 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 17-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3408	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGCACGGGGCCGUCGCCGA	UCGGCGACGGCCCCGUGCU	-2.4	-2.10 	-3.80 	2.00 	0.00 	8.00 	-49.7	27.70 	-6.10 	2.00 	II	78.90 	35.00 	28.48 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 18-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3409	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGCACGGGGCCGUCGCCG	CGGCGACGGCCCCGUGCUG	-2.4	-2.10 	-3.80 	1.00 	2.00 	8.00 	-49.4	21.50 	-1.70 	1.00 	II	84.20 	45.60 	35.09 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 19-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3410	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCAGCACGGGGCCGUCGCC	GGCGACGGCCCCGUGCUGC	-3.3	-3.40 	-3.50 	0.00 	2.00 	8.00 	-50.4	15.00 	7.50 	-1.00 	II	84.20 	29.70 	18.87 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 20-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3411	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGCAGCACGGGGCCGUCGC	GCGACGGCCCCGUGCUGCU	-3.4	-2.10 	-3.00 	2.00 	3.00 	8.00 	-49.2	37.90 	9.80 	2.00 	II	78.90 	49.10 	42.04 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 21-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3412	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGCAGCACGGGGCCGUCG	CGACGGCCCCGUGCUGCUG	-2.4	-2.10 	-3.00 	1.00 	0.00 	8.00 	-47.9	26.60 	0.70 	1.00 	II	78.90 	43.90 	40.44 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 22-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3413	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCAGCAGCACGGGGCCGUC	GACGGCCCCGUGCUGCUGC	-2.4	-3.40 	-3.00 	-1.00 	1.00 	8.00 	-48.9	15.20 	7.40 	0.00 	II	78.90 	28.80 	26.45 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 23-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3414	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCAGCAGCACGGGGCCGU	ACGGCCCCGUGCUGCUGCC	-2.2	-3.30 	-3.00 	0.00 	-1.00 	8.00 	-49.8	19.40 	-6.30 	1.00 	III	78.90 	27.00 	19.61 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 24-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3415	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGCAGCAGCACGGGGCCG	CGGCCCCGUGCUGCUGCCC	-2.4	-3.30 	-2.50 	1.00 	1.00 	8.00 	-50.9	8.70 	2.70 	-1.00 	II	84.20 	28.50 	14.04 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 25-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3416	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGGCAGCAGCACGGGGCC	GGCCCCGUGCUGCUGCCCG	-3.3	-2.40 	-1.70 	-2.00 	0.00 	7.00 	-50.9	15.50 	5.50 	-1.00 	II	84.20 	34.50 	21.14 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 26-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3417	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGGGCAGCAGCACGGGGC	GCCCCGUGCUGCUGCCCGA	-3.4	-2.40 	-0.60 	1.00 	3.00 	6.00 	-50	36.20 	2.50 	4.00 	II	78.90 	60.40 	36.52 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 27-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3418	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCGGGCAGCAGCACGGGG	CCCCGUGCUGCUGCCCGAC	-3.3	-2.20 	-0.60 	1.00 	2.00 	5.00 	-48.8	21.50 	4.70 	0.00 	II	78.90 	36.20 	48.29 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 28-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3419	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUCGGGCAGCAGCACGGG	CCCGUGCUGCUGCCCGACA	-3.3	-2.10 	-0.60 	1.00 	4.00 	5.00 	-47.6	27.80 	-2.40 	4.00 	II	73.70 	54.20 	38.55 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 29-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3420	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGUCGGGCAGCAGCACGG	CCGUGCUGCUGCCCGACAA	-3.3	-0.90 	-0.60 	0.00 	4.00 	5.00 	-45.2	49.50 	-4.90 	6.00 	II	68.40 	67.70 	77.17 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 30-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3421	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUUGUCGGGCAGCAGCACG	CGUGCUGCUGCCCGACAAC	-2.4	-2.20 	-0.60 	1.00 	2.00 	5.00 	-44.1	43.00 	4.70 	3.00 	II	68.40 	48.20 	79.88 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 31-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3422	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUUGUCGGGCAGCAGCAC	GUGCUGCUGCCCGACAACC	-2.2	-3.30 	-0.60 	1.00 	1.00 	5.00 	-45	30.60 	5.10 	1.00 	II	68.40 	30.70 	62.46 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 32-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3423	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGUUGUCGGGCAGCAGCA	UGCUGCUGCCCGACAACCA	-2.1	-2.10 	-0.60 	1.00 	0.00 	5.00 	-44.9	46.80 	-8.70 	5.00 	II	63.20 	53.00 	86.83 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 33-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3424	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGGUUGUCGGGCAGCAGC	GCUGCUGCCCGACAACCAC	-3.4	-2.20 	-0.60 	1.00 	1.00 	5.00 	-45	46.50 	9.80 	2.00 	II	68.40 	50.90 	72.87 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 34-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3425	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGUGGUUGUCGGGCAGCAG	CUGCUGCCCGACAACCACU	-2.1	-2.10 	2.00 	2.00 	3.00 	5.00 	-43.7	43.30 	0.00 	3.00 	Ib	63.20 	47.40 	54.20 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 35-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3426	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UAGUGGUUGUCGGGCAGCA	UGCUGCCCGACAACCACUA	-2.1	-1.30 	2.00 	1.00 	1.00 	5.00 	-42.9	47.30 	-6.10 	5.00 	II	57.90 	53.60 	56.72 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 36-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3427	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUAGUGGUUGUCGGGCAGC	GCUGCCCGACAACCACUAC	-3.4	-2.20 	2.00 	-1.00 	2.00 	5.00 	-43	35.70 	5.10 	3.00 	II	63.20 	49.80 	50.43 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 37-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3428	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUAGUGGUUGUCGGGCAG	CUGCCCGACAACCACUACC	-2.1	-3.30 	2.00 	-1.00 	1.00 	5.00 	-42.9	42.80 	5.40 	2.00 	II	63.20 	31.10 	51.29 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 38-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3429	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGGUAGUGGUUGUCGGGCA	UGCCCGACAACCACUACCU	-2.1	-2.10 	2.10 	3.00 	0.00 	5.00 	-42.9	47.50 	-1.10 	4.00 	II	57.90 	44.00 	61.02 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 39-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3430	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGGUAGUGGUUGUCGGGC	GCCCGACAACCACUACCUG	-3.4	-2.10 	3.00 	0.00 	2.00 	5.00 	-42.9	45.00 	10.80 	2.00 	II	63.20 	50.60 	52.66 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 40-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3431	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCAGGUAGUGGUUGUCGGG	CCCGACAACCACUACCUGA	-3.3	-2.40 	1.90 	-1.00 	4.00 	4.00 	-41.9	53.70 	3.00 	4.00 	Ib	57.90 	61.20 	81.93 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 41-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3432	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCAGGUAGUGGUUGUCGG	CCGACAACCACUACCUGAG	-3.3	-2.10 	1.90 	1.00 	2.00 	3.00 	-40.7	58.50 	0.30 	2.00 	II	57.90 	49.90 	90.38 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 42-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3433	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUCAGGUAGUGGUUGUCG	CGACAACCACUACCUGAGC	-2.4	-3.40 	1.90 	1.00 	2.00 	2.00 	-40.8	45.40 	5.00 	3.00 	II	57.90 	43.80 	46.17 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 43-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3434	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCUCAGGUAGUGGUUGUC	GACAACCACUACCUGAGCA	-2.4	-2.10 	1.90 	-1.00 	2.00 	2.00 	-40.5	61.40 	10.40 	6.00 	II	52.60 	52.90 	52.40 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 44-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3435	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGCUCAGGUAGUGGUUGU	ACAACCACUACCUGAGCAC	-2.2	-2.20 	1.90 	1.00 	-1.00 	2.00 	-40.3	53.80 	-1.70 	4.00 	III	52.60 	36.10 	62.88 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 45-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3436	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUGCUCAGGUAGUGGUUG	CAACCACUACCUGAGCACC	-2.1	-3.30 	1.90 	1.00 	1.00 	2.00 	-41.4	45.70 	2.30 	1.00 	II	57.90 	38.50 	40.65 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 46-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3437	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGUGCUCAGGUAGUGGUU	AACCACUACCUGAGCACCC	-0.9	-3.30 	1.90 	0.00 	-2.00 	3.00 	-42.6	45.70 	-3.50 	1.00 	III	57.90 	27.60 	79.56 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 47-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3438	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGGUGCUCAGGUAGUGGU	ACCACUACCUGAGCACCCA	-2.2	-2.10 	1.90 	1.00 	1.00 	3.00 	-43.8	58.00 	-4.00 	4.00 	II	57.90 	54.10 	97.99 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 48-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3439	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUGGGUGCUCAGGUAGUGG	CCACUACCUGAGCACCCAG	-3.3	-2.10 	-0.60 	-1.00 	1.00 	3.00 	-43.7	44.80 	2.60 	2.00 	II	63.20 	49.20 	75.84 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 49-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3440	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACUGGGUGCUCAGGUAGUG	CACUACCUGAGCACCCAGU	-2.1	-2.20 	-1.20 	2.00 	3.00 	3.00 	-42.6	36.80 	-2.40 	2.00 	II	57.90 	46.50 	51.33 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 50-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3441	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GACUGGGUGCUCAGGUAGU	ACUACCUGAGCACCCAGUC	-2.2	-2.40 	-1.20 	0.00 	-1.00 	3.00 	-42.9	38.40 	-8.90 	1.00 	III	57.90 	34.10 	36.88 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 51-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3442	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGACUGGGUGCUCAGGUAG	CUACCUGAGCACCCAGUCC	-2.1	-3.30 	-1.20 	1.00 	0.00 	3.00 	-44	40.00 	3.10 	2.00 	II	63.20 	37.40 	67.08 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 52-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3443	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGACUGGGUGCUCAGGUA	UACCUGAGCACCCAGUCCG	-1.3	-2.40 	-1.20 	0.00 	-4.00 	3.00 	-44.3	39.70 	-8.10 	0.00 	III	63.20 	26.60 	64.68 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 53-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3444	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCGGACUGGGUGCUCAGGU	ACCUGAGCACCCAGUCCGC	-2.2	-3.40 	-1.20 	2.00 	0.00 	4.00 	-46.4	32.50 	3.50 	0.00 	III	68.40 	33.90 	69.96 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 54-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3445	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCGGACUGGGUGCUCAGG	CCUGAGCACCCAGUCCGCC	-3.3	-3.30 	-1.20 	1.00 	0.00 	5.00 	-47.5	27.70 	3.00 	-2.00 	II	73.70 	28.50 	40.62 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 55-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3446	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGCGGACUGGGUGCUCAG	CUGAGCACCCAGUCCGCCC	-2.1	-3.30 	-0.90 	0.00 	0.00 	6.00 	-47.5	25.80 	0.00 	0.00 	II	73.70 	30.20 	41.95 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 56-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3447	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGGGCGGACUGGGUGCUCA	UGAGCACCCAGUCCGCCCU	-2.1	-2.10 	0.50 	2.00 	0.00 	6.00 	-47.5	36.20 	-1.40 	3.00 	II	68.40 	40.10 	49.79 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 57-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3448	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGGGCGGACUGGGUGCUC	GAGCACCCAGUCCGCCCUG	-2.4	-2.10 	-0.10 	-2.00 	0.00 	6.00 	-47.5	21.00 	5.40 	1.00 	II	73.70 	38.00 	44.88 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 58-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3449	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCAGGGCGGACUGGGUGCU	AGCACCCAGUCCGCCCUGA	-2.1	-2.40 	-1.20 	0.00 	2.00 	6.00 	-47.5	30.00 	-3.30 	3.00 	II	68.40 	47.10 	43.21 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 59-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3450	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCAGGGCGGACUGGGUGC	GCACCCAGUCCGCCCUGAG	-3.4	-2.10 	-1.20 	-1.00 	0.00 	6.00 	-47.5	37.30 	5.40 	2.00 	II	73.70 	45.50 	64.82 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 60-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3451	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUCAGGGCGGACUGGGUG	CACCCAGUCCGCCCUGAGC	-2.1	-3.40 	-1.20 	2.00 	2.00 	6.00 	-47.5	28.50 	7.40 	1.00 	II	73.70 	34.50 	51.63 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 61-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3452	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCUCAGGGCGGACUGGGU	ACCCAGUCCGCCCUGAGCA	-2.2	-2.10 	-1.20 	1.00 	1.00 	6.00 	-47.5	41.80 	-6.60 	3.00 	II	68.40 	43.30 	51.85 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 62-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3453	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGCUCAGGGCGGACUGGG	CCCAGUCCGCCCUGAGCAA	-3.3	-0.90 	-1.20 	2.00 	4.00 	6.00 	-46.2	43.10 	-2.40 	4.00 	Ib	68.40 	64.40 	34.08 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 63-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3454	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUUGCUCAGGGCGGACUGG	CCAGUCCGCCCUGAGCAAA	-3.3	-0.90 	-1.20 	0.00 	4.00 	6.00 	-43.8	49.10 	-2.40 	4.00 	Ib	63.20 	60.00 	73.77 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 64-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3455	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUUGCUCAGGGCGGACUG	CAGUCCGCCCUGAGCAAAG	-2.1	-2.10 	-0.90 	-1.00 	1.00 	6.00 	-42.6	51.90 	0.40 	3.00 	II	63.20 	46.20 	77.81 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 10-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3456	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCUUUGCUCAGGGCGGACU	AGUCCGCCCUGAGCAAAGA	-2.1	-2.40 	-0.80 	1.00 	1.00 	6.00 	-42.9	61.90 	-4.00 	6.00 	II	57.90 	56.20 	83.90 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 11-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3457	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCUUUGCUCAGGGCGGAC	GUCCGCCCUGAGCAAAGAC	-2.2	-2.20 	-0.80 	-1.00 	1.00 	6.00 	-43	46.00 	9.70 	3.00 	II	63.20 	38.40 	80.58 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 12-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3458	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUCUUUGCUCAGGGCGGA	UCCGCCCUGAGCAAAGACC	-2.4	-3.30 	-0.80 	2.00 	-1.00 	6.00 	-44.1	32.90 	-6.10 	2.00 	II	63.20 	25.90 	54.93 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 13-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3459	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGUCUUUGCUCAGGGCGG	CCGCCCUGAGCAAAGACCC	-3.3	-3.30 	-0.80 	1.00 	1.00 	6.00 	-45	41.70 	-4.90 	1.00 	II	68.40 	40.30 	44.99 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 14-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3460	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGGGUCUUUGCUCAGGGCG	CGCCCUGAGCAAAGACCCC	-2.4	-3.30 	-0.80 	1.00 	1.00 	5.00 	-45	46.90 	3.10 	1.00 	II	68.40 	44.70 	59.56 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 15-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3461	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGGGGUCUUUGCUCAGGGC	GCCCUGAGCAAAGACCCCA	-3.4	-2.10 	-0.50 	0.00 	3.00 	4.00 	-44.7	53.00 	5.10 	2.00 	II	63.20 	56.50 	52.41 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 16-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3462	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGGGGUCUUUGCUCAGGG	CCCUGAGCAAAGACCCCAA	-3.3	-0.90 	0.50 	1.00 	5.00 	4.00 	-42.2	58.60 	4.70 	4.00 	II	57.90 	70.90 	80.40 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 17-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3463	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUUGGGGUCUUUGCUCAGG	CCUGAGCAAAGACCCCAAC	-3.3	-2.20 	0.50 	1.00 	2.00 	4.00 	-41.1	47.60 	3.00 	2.00 	II	57.90 	47.70 	58.79 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 18-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3464	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUUGGGGUCUUUGCUCAG	CUGAGCAAAGACCCCAACG	-2.1	-2.40 	0.80 	-2.00 	0.00 	4.00 	-40.2	46.40 	1.00 	2.00 	II	57.90 	37.20 	50.17 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 19-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3465	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGUUGGGGUCUUUGCUCA	UGAGCAAAGACCCCAACGA	-2.1	-2.40 	2.70 	1.00 	0.00 	4.00 	-40.5	65.10 	-0.80 	7.00 	II	52.60 	53.10 	81.08 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 20-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3466	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCGUUGGGGUCUUUGCUC	GAGCAAAGACCCCAACGAG	-2.4	-2.10 		0.00 	1.00 	4.00 	-40.5	56.50 	5.40 	2.00 	II	57.90 	53.40 	79.64 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 21-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3467	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCUCGUUGGGGUCUUUGCU	AGCAAAGACCCCAACGAGA	-2.1	-2.40 		3.00 	3.00 	4.00 	-40.5	68.70 	3.80 	5.00 	II	52.60 	57.90 	82.37 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 22-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3468	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUCUCGUUGGGGUCUUUGC	GCAAAGACCCCAACGAGAA	-3.4	-0.90 		2.00 	3.00 	4.00 	-39.3	82.90 	7.40 	7.00 	Ib	52.60 	70.20 	92.10 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 23-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3469	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUCUCGUUGGGGUCUUUG	CAAAGACCCCAACGAGAAG	-2.1	-2.10 		0.00 	1.00 	4.00 	-38	64.50 	0.30 	5.00 	II	52.60 	51.40 	71.51 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 24-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3470	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUUCUCGUUGGGGUCUUU	AAAGACCCCAACGAGAAGC	-0.9	-3.40 		-1.00 	-2.00 	4.00 	-39.3	48.80 	-4.00 	3.00 	III	52.60 	23.00 	14.62 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 25-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3471	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGCUUCUCGUUGGGGUCUU	AAGACCCCAACGAGAAGCG	-0.9	-2.40 	1.40 	0.00 	-3.00 	4.00 	-40.8	44.20 	-6.00 	2.00 	III	57.90 	29.10 	11.33 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 26-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3472	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCGCUUCUCGUUGGGGUCU	AGACCCCAACGAGAAGCGC	-2.1	-3.40 	1.00 	1.00 	-2.00 	4.00 	-43.3	39.70 	-1.00 	1.00 	III	63.20 	31.50 	17.22 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 27-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3473	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGCGCUUCUCGUUGGGGUC	GACCCCAACGAGAAGCGCG	-2.4	-2.40 	1.00 	-1.00 	-1.00 	5.00 	-43.6	49.30 	8.40 	0.00 	II	68.40 	38.10 	31.83 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 28-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3474	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGCGCUUCUCGUUGGGGU	ACCCCAACGAGAAGCGCGA	-2.2	-2.40 	0.60 	3.00 	2.00 	5.00 	-43.6	55.20 	-4.00 	3.00 	II	63.20 	51.00 	84.64 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 29-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3475	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUCGCGCUUCUCGUUGGGG	CCCCAACGAGAAGCGCGAU	-3.3	-1.10 	0.60 	1.00 	5.00 	5.00 	-42.5	48.20 	0.00 	2.00 	II	63.20 	53.50 	84.64 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 30-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3476	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAUCGCGCUUCUCGUUGGG	CCCAACGAGAAGCGCGAUC	-3.3	-2.40 	0.60 	0.00 	3.00 	5.00 	-41.6	43.80 	3.10 	2.00 	II	63.20 	36.90 	74.41 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 31-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3477	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGAUCGCGCUUCUCGUUGG	CCAACGAGAAGCGCGAUCA	-3.3	-2.10 	0.60 	1.00 	3.00 	5.00 	-40.4	63.80 	-2.40 	5.00 	II	57.90 	60.10 	83.93 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 32-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3478	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGAUCGCGCUUCUCGUUG	CAACGAGAAGCGCGAUCAC	-2.1	-2.20 	0.60 	1.00 	1.00 	5.00 	-39.3	61.10 	7.80 	4.00 	II	57.90 	49.60 	78.72 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 33-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3479	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUGAUCGCGCUUCUCGUU	AACGAGAAGCGCGAUCACA	-0.9	-2.10 	0.60 	1.00 	1.00 	5.00 	-39.3	62.30 	-0.60 	5.00 	II	52.60 	52.20 	77.66 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 34-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3480	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUGUGAUCGCGCUUCUCGU	ACGAGAAGCGCGAUCACAU	-2.2	-1.10 	0.60 	3.00 	3.00 	5.00 	-39.5	65.80 	-4.00 	5.00 	II	52.60 	55.10 	83.23 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 35-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3481	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAUGUGAUCGCGCUUCUCG	CGAGAAGCGCGAUCACAUG	-2.4	-2.10 	0.90 	0.00 	2.00 	5.00 	-39.4	62.70 	0.40 	4.00 	II	57.90 	58.50 	82.60 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 36-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3482	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCAUGUGAUCGCGCUUCUC	GAGAAGCGCGAUCACAUGG	-2.4	-3.30 	1.20 	-1.00 	0.00 	5.00 	-40.3	46.90 	3.10 	2.00 	II	57.90 	43.90 	58.75 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 37-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3483	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACCAUGUGAUCGCGCUUCU	AGAAGCGCGAUCACAUGGU	-2.1	-2.20 	1.00 	1.00 	0.00 	5.00 	-40.1	54.90 	-3.90 	6.00 	II	52.60 	45.30 	47.59 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 38-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3484	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GACCAUGUGAUCGCGCUUC	GAAGCGCGAUCACAUGGUC	-2.4	-2.40 	-1.80 	2.00 	0.00 	5.00 	-40.4	50.10 	7.80 	3.00 	II	57.90 	49.20 	61.12 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 39-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3485	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGACCAUGUGAUCGCGCUU	AAGCGCGAUCACAUGGUCC	-0.9	-3.30 	-2.60 	1.00 	-1.00 	5.00 	-41.3	42.60 	3.80 	0.00 	III	57.90 	30.90 	43.28 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 40-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3486	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGGACCAUGUGAUCGCGCU	AGCGCGAUCACAUGGUCCU	-2.1	-2.10 	-2.60 	2.00 	1.00 	5.00 	-42.5	56.50 	-4.00 	2.00 	II	57.90 	48.20 	61.73 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 41-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3487	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGGACCAUGUGAUCGCGC	GCGCGAUCACAUGGUCCUG	-3.4	-2.10 	-2.60 	-1.00 	2.00 	5.00 	-42.5	41.00 	7.80 	0.00 	II	63.20 	50.70 	47.42 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 42-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3488	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCAGGACCAUGUGAUCGCG	CGCGAUCACAUGGUCCUGC	-2.4	-3.40 	-2.60 	0.00 	2.00 	4.00 	-42.5	32.30 	3.00 	-1.00 	II	63.20 	34.60 	29.11 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 43-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3489	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGCAGGACCAUGUGAUCGC	GCGAUCACAUGGUCCUGCU	-3.4	-2.10 	-2.60 	1.00 	3.00 	3.00 	-42.2	50.10 	7.40 	3.00 	II	57.90 	56.30 	38.65 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 44-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3490	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGCAGGACCAUGUGAUCG	CGAUCACAUGGUCCUGCUG	-2.4	-2.10 	-2.60 	1.00 	1.00 	2.00 	-40.9	52.90 	8.40 	2.00 	II	57.90 	53.40 	50.54 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 45-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3491	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCAGCAGGACCAUGUGAUC	GAUCACAUGGUCCUGCUGG	-2.4	-3.30 	-2.30 	-2.00 	-1.00 	2.00 	-41.8	36.60 	3.10 	0.00 	II	57.90 	36.50 	62.90 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 46-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3492	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCCAGCAGGACCAUGUGAU	AUCACAUGGUCCUGCUGGA	-1.1	-2.40 	-1.80 	1.00 	1.00 	2.00 	-41.8	57.00 	-8.90 	5.00 	II	52.60 	50.80 	76.45 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 47-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3493	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCCAGCAGGACCAUGUGA	UCACAUGGUCCUGCUGGAG	-2.4	-2.10 	-2.60 	1.00 	-3.00 	2.00 	-42.8	39.50 	-0.70 	1.00 	III	57.90 	33.70 	36.24 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 48-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3494	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACUCCAGCAGGACCAUGUG	CACAUGGUCCUGCUGGAGU	-2.1	-2.20 	-2.60 	2.00 	3.00 	2.00 	-42.6	46.00 	2.40 	2.00 	II	57.90 	45.40 	78.82 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 49-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3495	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AACUCCAGCAGGACCAUGU	ACAUGGUCCUGCUGGAGUU	-2.2	-0.90 	-2.60 	2.00 	1.00 	2.00 	-41.4	68.00 	-6.60 	6.00 	II	52.60 	57.20 	56.39 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 50-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3496	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAACUCCAGCAGGACCAUG	CAUGGUCCUGCUGGAGUUC	-2.1	-2.40 	-2.60 	2.00 	1.00 	2.00 	-41.6	42.10 	2.30 	1.00 	II	57.90 	42.60 	71.90 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 51-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3497	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGAACUCCAGCAGGACCAU	AUGGUCCUGCUGGAGUUCG	-1.1	-2.40 	-2.60 	-2.00 	-2.00 	2.00 	-41.9	33.30 	-8.30 	0.00 	III	57.90 	27.30 	30.83 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 52-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3498	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACGAACUCCAGCAGGACCA	UGGUCCUGCUGGAGUUCGU	-2.1	-2.20 	-2.60 	2.00 	0.00 	2.00 	-43	49.10 	-6.00 	4.00 	II	57.90 	49.70 	86.15 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 53-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3499	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CACGAACUCCAGCAGGACC	GGUCCUGCUGGAGUUCGUG	-3.3	-2.10 	-2.60 	0.00 	0.00 	2.00 	-43	45.50 	12.80 	0.00 	II	63.20 	46.70 	74.54 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 54-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3500	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCACGAACUCCAGCAGGAC	GUCCUGCUGGAGUUCGUGA	-2.2	-2.40 	-2.60 	1.00 	3.00 	2.00 	-42.1	38.60 	5.10 	3.00 	Ib	57.90 	54.00 	44.11 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 55-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3501	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCACGAACUCCAGCAGGA	UCCUGCUGGAGUUCGUGAC	-2.4	-2.20 	-2.40 	0.00 	-1.00 	2.00 	-42.1	39.10 	-11.00 	1.00 	III	57.90 	34.50 	85.24 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 56-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3502	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGUCACGAACUCCAGCAGG	CCUGCUGGAGUUCGUGACC	-3.3	-3.30 	-0.80 	1.00 	1.00 	2.00 	-43	37.20 	2.80 	2.00 	II	63.20 	39.60 	52.67 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 57-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3503	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGUCACGAACUCCAGCAG	CUGCUGGAGUUCGUGACCG	-2.1	-2.40 	0.30 	-1.00 	-1.00 	3.00 	-42.1	45.20 	1.10 	1.00 	II	63.20 	32.00 	75.73 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 58-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3504	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCGGUCACGAACUCCAGCA	UGCUGGAGUUCGUGACCGC	-2.1	-3.40 	0.30 	1.00 	-2.00 	4.00 	-43.4	45.10 	-3.80 	2.00 	III	63.20 	37.80 	36.89 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 59-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3505	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCGGUCACGAACUCCAGC	GCUGGAGUUCGUGACCGCC	-3.4	-3.30 	0.30 	1.00 	1.00 	5.00 	-44.6	39.10 	14.40 	-1.00 	II	68.40 	39.50 	30.95 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 60-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3506	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGCGGUCACGAACUCCAG	CUGGAGUUCGUGACCGCCG	-2.1	-2.40 	0.30 	1.00 	-1.00 	6.00 	-43.6	31.70 	-4.60 	0.00 	II	68.40 	36.70 	39.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 61-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3507	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCGGCGGUCACGAACUCCA	UGGAGUUCGUGACCGCCGC	-2.1	-3.40 	-0.10 	1.00 	-2.00 	7.00 	-44.9	30.00 	-8.70 	1.00 	III	68.40 	35.10 	32.90 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 62-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3508	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GGCGGCGGUCACGAACUCC	GGAGUUCGUGACCGCCGCC	-3.3	-3.30 	-0.90 	0.00 	0.00 	8.00 	-46.1	26.40 	7.40 	1.00 	II	73.70 	38.20 	40.14 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 63-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3509	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGGCGGCGGUCACGAACUC	GAGUUCGUGACCGCCGCCG	-2.4	-2.40 	-0.90 	0.00 	-1.00 	9.00 	-45.2	13.50 	5.50 	-1.00 	II	73.70 	28.60 	28.30 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 64-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3510	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCGGCGGCGGUCACGAACU	AGUUCGUGACCGCCGCCGG	-2.1	-3.30 	-1.30 	0.00 	-3.00 	10.00 	-46.1	23.50 	-10.90 	-1.00 	III	73.70 	38.20 	30.00 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 10-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3511	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCCGGCGGCGGUCACGAAC	GUUCGUGACCGCCGCCGGG	-2.2	-3.30 	-2.60 	0.00 	-2.00 	11.00 	-47.3	26.80 	10.80 	-1.00 	II	78.90 	33.10 	33.47 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 11-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3512	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCCCGGCGGCGGUCACGAA	UUCGUGACCGCCGCCGGGA	-0.9	-2.40 	-2.60 	2.00 	-1.00 	11.00 	-47.5	24.80 	-3.80 	1.00 	II	73.70 	29.80 	29.13 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 12-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3513	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUCCCGGCGGCGGUCACGA	UCGUGACCGCCGCCGGGAU	-2.4	-1.10 	-2.60 	3.00 	2.00 	11.00 	-47.7	30.60 	-3.80 	2.00 	II	73.70 	43.90 	34.74 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 13-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3514	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAUCCCGGCGGCGGUCACG	CGUGACCGCCGCCGGGAUC	-2.4	-2.40 	-2.60 	1.00 	2.00 	11.00 	-47.7	29.20 	-2.40 	1.00 	II	78.90 	39.20 	25.27 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 14-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3515	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGAUCCCGGCGGCGGUCAC	GUGACCGCCGCCGGGAUCA	-2.2	-2.10 	-2.60 	0.00 	3.00 	11.00 	-47.4	42.50 	9.80 	3.00 	II	73.70 	44.60 	69.59 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 15-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3516	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUGAUCCCGGCGGCGGUCA	UGACCGCCGCCGGGAUCAC	-2.1	-2.20 	-2.60 	1.00 	-2.00 	11.00 	-47.4	33.60 	-6.10 	2.00 	III	73.70 	32.30 	14.39 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 16-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3517	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGUGAUCCCGGCGGCGGUC	GACCGCCGCCGGGAUCACU	-2.4	-2.10 	-2.60 	1.00 	3.00 	11.00 	-47.4	32.00 	7.80 	2.00 	Ib	73.70 	45.10 	30.53 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 17-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3518	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAGUGAUCCCGGCGGCGGU	ACCGCCGCCGGGAUCACUC	-2.2	-2.40 	-2.60 	1.00 	0.00 	11.00 	-47.4	32.30 	-1.60 	1.00 	II	73.70 	30.20 	57.62 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 18-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3519	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGAGUGAUCCCGGCGGCGG	CCGCCGCCGGGAUCACUCU	-3.3	-2.10 	-2.10 	2.00 	4.00 	11.00 	-47.3	35.00 	-2.40 	3.00 	Ib	73.70 	54.30 	26.93 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 19-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3520	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GAGAGUGAUCCCGGCGGCG	CGCCGCCGGGAUCACUCUC	-2.4	-2.40 	-1.00 	-1.00 	2.00 	10.00 	-46.4	26.80 	-4.70 	1.00 	II	73.70 	42.70 	32.57 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 20-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3521	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGAGAGUGAUCCCGGCGGC	GCCGCCGGGAUCACUCUCG	-3.4	-2.40 	-3.40 	-1.00 	1.00 	9.00 	-46.4	25.20 	5.80 	1.00 	II	73.70 	38.90 	47.36 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 21-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3522	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCGAGAGUGAUCCCGGCGG	CCGCCGGGAUCACUCUCGG	-3.3	-3.30 	-6.60 	0.00 	0.00 	8.00 	-46.3	30.70 	-2.00 	1.00 	II	73.70 	42.60 	29.72 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 22-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3523	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCCGAGAGUGAUCCCGGCG	CGCCGGGAUCACUCUCGGC	-2.4	-3.40 	-10.10 	0.00 	1.00 	7.00 	-46.4	32.10 	10.40 	0.00 	II	73.70 	40.30 	31.46 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 23-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3524	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGCCGAGAGUGAUCCCGGC	GCCGGGAUCACUCUCGGCA	-3.4	-2.10 	-11.50 	2.00 	3.00 	6.00 	-46.1	38.80 	7.40 	2.00 	II	68.40 	53.10 	15.45 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 24-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3525	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AUGCCGAGAGUGAUCCCGG	CCGGGAUCACUCUCGGCAU	-3.3	-1.10 	-6.80 	2.00 	5.00 	5.00 	-43.8	45.60 	-0.30 	2.00 	II	63.20 	63.20 	25.13 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 25-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3526	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAUGCCGAGAGUGAUCCCG	CGGGAUCACUCUCGGCAUG	-2.4	-2.10 	-4.40 	0.00 	2.00 	4.00 	-42.6	49.70 	1.00 	2.00 	II	63.20 	52.90 	31.76 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 26-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3527	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CCAUGCCGAGAGUGAUCCC	GGGAUCACUCUCGGCAUGG	-3.3	-3.30 	-2.40 	-2.00 	1.00 	4.00 	-43.5	41.40 	7.70 	0.00 	II	63.20 	39.50 	24.37 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 27-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3528	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCCAUGCCGAGAGUGAUCC	GGAUCACUCUCGGCAUGGA	-3.3	-2.40 	1.00 	1.00 	3.00 	4.00 	-42.6	62.30 	9.70 	6.00 	II	57.90 	65.60 	62.83 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 28-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3529	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUCCAUGCCGAGAGUGAUC	GAUCACUCUCGGCAUGGAC	-2.4	-2.20 	0.80 	1.00 	0.00 	4.00 	-41.5	45.60 	9.80 	2.00 	II	57.90 	44.00 	26.06 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 29-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3530	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CGUCCAUGCCGAGAGUGAU	AUCACUCUCGGCAUGGACG	-1.1	-2.40 	0.80 	2.00 	-2.00 	4.00 	-41.5	40.60 	-6.00 	1.00 	III	57.90 	28.00 	21.86 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 30-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3531	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UCGUCCAUGCCGAGAGUGA	UCACUCUCGGCAUGGACGA	-2.4	-2.40 	-0.90 	0.00 	-1.00 	4.00 	-42.8	48.20 	-8.70 	5.00 	II	57.90 	44.60 	11.37 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 31-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3532	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUCGUCCAUGCCGAGAGUG	CACUCUCGGCAUGGACGAG	-2.1	-2.10 	-3.50 	-1.00 	0.00 	4.00 	-42.5	38.90 	-6.70 	0.00 	II	63.20 	43.50 	25.11 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 32-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3533	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GCUCGUCCAUGCCGAGAGU	ACUCUCGGCAUGGACGAGC	-2.2	-3.40 	-3.90 	0.00 	-1.00 	4.00 	-43.8	32.10 	-3.90 	-1.00 	III	63.20 	21.50 	9.14 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 33-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3534	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	AGCUCGUCCAUGCCGAGAG	CUCUCGGCAUGGACGAGCU	-2.1	-2.10 	-3.90 	3.00 	2.00 	4.00 	-43.7	56.60 	2.40 	3.00 	II	63.20 	51.30 	60.36 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 34-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3535	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CAGCUCGUCCAUGCCGAGA	UCUCGGCAUGGACGAGCUG	-2.4	-2.10 	-3.90 	1.00 	-3.00 	4.00 	-43.7	45.90 	-0.40 	1.00 	III	63.20 	37.10 	83.09 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 35-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3536	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACAGCUCGUCCAUGCCGAG	CUCGGCAUGGACGAGCUGU	-2.1	-2.20 	-3.80 	0.00 	3.00 	4.00 	-43.5	37.70 	-2.40 	1.00 	II	63.20 	42.60 	33.35 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 36-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3537	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UACAGCUCGUCCAUGCCGA	UCGGCAUGGACGAGCUGUA	-2.4	-1.30 	-3.50 	2.00 	2.00 	4.00 	-42.7	57.50 	-8.70 	4.00 	II	57.90 	56.70 	66.79 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 37-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3538	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	GUACAGCUCGUCCAUGCCG	CGGCAUGGACGAGCUGUAC	-2.4	-2.20 	-3.40 	2.00 	3.00 	4.00 	-42.5	41.10 	2.80 	1.00 	II	63.20 	51.50 	56.75 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 38-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3539	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UGUACAGCUCGUCCAUGCC	GGCAUGGACGAGCUGUACA	-3.3	-2.10 	-1.60 	0.00 	4.00 	3.00 	-42.2	66.00 	12.80 	4.00 	Ib	57.90 	62.40 	73.66 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 39-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3540	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUGUACAGCUCGUCCAUGC	GCAUGGACGAGCUGUACAA	-3.4	-0.90 	-0.50 	1.00 	3.00 	2.00 	-39.8	73.70 	7.80 	8.00 	Ia	52.60 	72.50 	76.68 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 40-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3541	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	CUUGUACAGCUCGUCCAUG	CAUGGACGAGCUGUACAAG	-2.1	-2.10 	0.60 	0.00 	1.00 	2.00 	-38.5	49.50 	-2.00 	3.00 	II	52.60 	48.00 	53.54 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 41-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3542	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	ACUUGUACAGCUCGUCCAU	AUGGACGAGCUGUACAAGU	-1.1	-2.20 	-0.20 	0.00 	1.00 	2.00 	-38.6	59.40 	-0.90 	5.00 	II	47.40 	42.20 	68.43 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 42-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3543	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UACUUGUACAGCUCGUCCA	UGGACGAGCUGUACAAGUA	-2.1	-1.30 	-0.90 	2.00 	1.00 	2.00 	-38.8	80.60 	-6.10 	8.00 	II	47.40 	66.20 	82.90 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 43-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26
Si3544	NZ_CP024869	Takayuki Katoh	EGFP 	2315885245	UUACUUGUACAGCUCGUCC	GGACGAGCUGUACAAGUAA	-3.3	-0.90 	-0.90 	1.00 	5.00 	2.00 	-37.6	82.80 	14.40 	8.00 	Ia	47.40 	79.60 	83.47 	Flow cytometry and immunofluorescence	HeLa	150nM	72h	"Specific residues at every third position of siRNA shape its efficient RNAi activity.
shape its efficient RNAi activity"	17259216	2007	Small interfering RNA (siRNA) induces sequence-specific post-transcriptional gene silencing in mammalian cells. Different efficacy of each siRNA is considered to result from sequence preference by protein components in RNAi. To obtain mechanistic insight into siRNA functionality, here we describe a complete data set of siRNA activities targeting all possible position of a single mRNA in human cells. Seven hundred and two siRNAs covering open reading frame of enhanced green fluorescent protein mRNA ( 720 bases) were examined with minimized error factors. The most important finding is that specific residues at every third position of siRNAs greatly influence its RNAi activity; the optimized base composition at positions 3n + 1 (4,7,10,13,16,19) in siRNAs have positive effects on the activity, which can explain the waving siRNA activity with 3 nucleotides (nt) periodicity in the sequential positions of mRNAs. Since there was an obvious correlation between siRNA activity and its binding affinity to TRBP, a partner protein of human Dicer, the 3-nt periodicity might correlate with the affinity to TRBP. As an algorithm (‘siExplorer’) developed by this observation successfully calculated the activities of siRNAs targeting endogenous human genes, the 44-nt periodicity provides a new aspect unveiling siRNA functionality.	Nucleic Acids Research	2007/1/26